Repository navigation
Expand file tree
/
Copy pathindex.html
More file actions
397 lines (366 loc) · 26.9 KB
/
Copy pathindex.html
File metadata and controls
397 lines (366 loc) · 26.9 KB
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
<!doctype html>
<html lang="en">
<head>
<meta charset="utf-8">
<meta name="viewport" content="width=device-width, initial-scale=1">
<title>Amr Amer | Machine Learning Engineer, Computer Vision</title>
<meta name="description" content="Amr Amer, Machine Learning Engineer with an M.Sc. in Visual Computing from Saarland University and experience at DFKI. Computer vision, multimodal generative models, data pipelines and MLOps.">
<meta name="author" content="Amr Amer">
<meta property="og:title" content="Amr Amer | Machine Learning Engineer">
<meta property="og:description" content="Computer vision, multimodal generative models, data pipelines and MLOps. Selected projects and CV.">
<meta property="og:type" content="website">
<meta property="og:url" content="https://amramer.github.io/">
<meta property="og:image" content="https://amramer.github.io/assets/img/badminton-tracking-poster.webp">
<meta name="theme-color" content="#f4f6f8" media="(prefers-color-scheme: light)">
<meta name="theme-color" content="#0f1317" media="(prefers-color-scheme: dark)">
<link rel="icon" href="data:image/svg+xml,%3Csvg xmlns='http://www.w3.org/2000/svg' viewBox='0 0 32 32'%3E%3Crect width='32' height='32' rx='7' fill='%230b6e69'/%3E%3Cpath d='M8 13V8h5M19 8h5v5M24 19v5h-5M13 24H8v-5' fill='none' stroke='%23fff' stroke-width='2.4'/%3E%3C/svg%3E">
<link rel="preload" href="assets/fonts/ibm-plex-sans-latin-400-normal.woff2" as="font" type="font/woff2" crossorigin>
<link rel="preload" href="assets/fonts/ibm-plex-sans-latin-600-normal.woff2" as="font" type="font/woff2" crossorigin>
<link rel="stylesheet" href="assets/css/style.css">
<script>
try { const t = localStorage.getItem("theme"); if (t) document.documentElement.dataset.theme = t; } catch (e) {}
</script>
</head>
<body>
<a class="skip" href="#main">Skip to content</a>
<header class="nav">
<div class="nav-inner">
<a class="nav-name" href="#top">Amr Amer</a>
<nav aria-label="Sections">
<a href="#projects">Projects</a>
<a href="#skills">Skills</a>
<a href="#beyond">Beyond work</a>
<a href="#contact">Contact</a>
</nav>
<div class="nav-actions">
<button class="theme-btn" type="button" aria-label="Switch between light and dark theme">
<svg class="i-sun" viewBox="0 0 24 24" aria-hidden="true"><circle cx="12" cy="12" r="4.5"/><path d="M12 2v2.5M12 19.5V22M2 12h2.5M19.5 12H22M4.9 4.9l1.8 1.8M17.3 17.3l1.8 1.8M4.9 19.1l1.8-1.8M17.3 6.7l1.8-1.8"/></svg>
<svg class="i-moon" viewBox="0 0 24 24" aria-hidden="true"><path d="M20 14.5A8 8 0 0 1 9.5 4a8 8 0 1 0 10.5 10.5Z"/></svg>
</button>
<a class="btn btn-small" href="assets/Resume_Amr_Amer.pdf" download="Amr_Amer_CV_ML_Engineer.pdf">Download CV</a>
</div>
</div>
</header>
<main id="main">
<!-- ============ HERO ============ -->
<section class="hero wrap" id="top">
<div class="hero-text">
<h1>Amr Amer</h1>
<p class="role">Machine Learning Engineer | Computer Vision | Deep Learning</p>
<p>Machine Learning Engineer with an M.Sc. in Visual Computing from Saarland University and experience at DFKI (German Research Center for Artificial Intelligence). I build production-ready ML systems end-to-end, from versioned, quality-checked data pipelines and distributed GPU training to multimodal generative models, high-performance C++ inference and deployment, with a focus on computer vision.</p>
<p>On the data and MLOps side, I build reproducible pipelines with DVC, Airflow, Spark and MLflow, add automated data-quality gates, and ship models with Docker, Kubernetes and CI/CD on GitHub Actions.</p>
<p class="seeking">Currently seeking roles in Machine Learning Engineering and Computer Vision.</p>
<div class="hero-actions">
<a class="btn" href="mailto:amribrahim.amer@gmail.com">Email</a>
<a class="btn btn-ghost" href="https://www.linkedin.com/in/amr-amer-2023-cs/" target="_blank" rel="noopener">LinkedIn</a>
<a class="btn btn-ghost" href="https://github.com/amramer" target="_blank" rel="noopener">GitHub</a>
<a class="btn btn-ghost" href="assets/Resume_Amr_Amer.pdf" download="Amr_Amer_CV_ML_Engineer.pdf">Download CV</a>
</div>
<div class="hero-meta">
<span class="otw"><span class="otw-dot" aria-hidden="true"></span>Open to work</span>
<a class="phone" href="tel:+491626667294">+49 162 6667294</a>
<a class="wa" href="https://wa.me/491626667294" target="_blank" rel="noopener" aria-label="Message on WhatsApp">WhatsApp</a>
</div>
</div>
<figure class="detect" aria-label="Portrait of Amr Amer">
<span class="bb-tag" aria-hidden="true">ml_engineer</span>
<picture>
<source srcset="assets/img/profile.webp" type="image/webp">
<img src="assets/img/profile.jpg" alt="Portrait of Amr Amer" width="480" height="743">
</picture>
<span class="corner tl" aria-hidden="true"></span><span class="corner tr" aria-hidden="true"></span>
<span class="corner bl" aria-hidden="true"></span><span class="corner br" aria-hidden="true"></span>
</figure>
</section>
<!-- ============ WHAT I DO ============ -->
<section class="wrap what" aria-labelledby="h-what">
<h2 id="h-what">What I do</h2>
<ul class="what-list">
<li>Design and train deep learning models for real-world computer vision problems</li>
<li>Build <strong>end-to-end ML pipelines</strong>: data → training → evaluation → deployment and visualization</li>
<li>Build <strong>reproducible data pipelines</strong> with dataset versioning, data-quality gates and CI/CD</li>
<li>Optimize models through <strong>systematic experimentation</strong> and hyperparameter tuning</li>
<li>Develop interactive demos and dashboards for analysis and presentation</li>
</ul>
</section>
<!-- ============ FEATURED ============ -->
<section class="wrap" id="projects" aria-labelledby="h-featured">
<h2 id="h-featured">Featured projects</h2>
<!-- Thesis -->
<article class="project" aria-labelledby="p-thesis">
<header class="project-head">
<p class="project-kind">Master's thesis</p>
<h3 id="p-thesis">Personality-Aware Non-verbal Behavior Generation in Dyadic Interactions</h3>
<p class="project-lede">Multimodal generative model for generating <strong>realistic listener avatars in dyadic conversations</strong>, conditioned on <strong>personality traits</strong>. The model predicts <strong>facial expressions, head motion and upper-body gestures</strong> of a listener from the speaker's <strong>audio and motion signals</strong>.</p>
</header>
<figure class="media media-dark media-native">
<video class="autoplay" muted loop playsinline preload="none" poster="assets/img/thesis-avatars-poster.webp" width="700" height="240" aria-label="Generated listener avatars with extrovert, neutral and introvert personalities">
<source src="assets/media/thesis-avatars.mp4" type="video/mp4">
</video>
</figure>
<div class="project-body">
<div>
<h4>Problem</h4>
<p>Human conversations rely heavily on <strong>non-verbal cues</strong> such as facial expressions, eye gaze and body movement. Most prior work does not model how these behaviors vary with <strong>personality traits</strong>.</p>
<h4>Approach</h4>
<p>Developed a <strong>multimodal generative architecture</strong> combining:</p>
<ul>
<li><strong>Transformer encoder</strong> for temporal modeling</li>
<li><strong>VQ-VAE</strong> for quantized motion representation learning</li>
<li><strong>SMPL-X body parameters</strong> for realistic face and upper-body avatar generation</li>
<li>Integration of <strong>audio, facial motion and body gesture signals</strong></li>
</ul>
<p>The model generates <strong>personality-aware listener behavior sequences</strong> in response to a speaker.</p>
</div>
<div>
<h4>Results</h4>
<dl class="metrics">
<div><dt>Face FID</dt><dd>6.15</dd></div>
<div><dt>Face P-FID</dt><dd>10.31</dd></div>
<div><dt>Upper-body FID</dt><dd>43.16</dd></div>
<div><dt>Upper-body P-FID</dt><dd>87.73</dd></div>
</dl>
<p class="small">User study: participants correctly identified <strong>extroverted vs. introverted avatars in 86% of cases</strong>, and <strong>71% preferred</strong> the personality-aware model over the personality-agnostic baseline.</p>
<h4>Tech stack</h4>
<ul class="chips">
<li>PyTorch</li><li>Transformers</li><li>VQ-VAE</li><li>SMPL-X (PIXIE)</li><li>Librosa</li><li>OpenCV</li><li>CUDA</li><li>TensorBoard</li><li>SLURM</li><li>Enroot</li><li>Multi-GPU training (A100)</li>
</ul>
</div>
</div>
<ul class="links">
<li><a class="btn" href="https://amramer.github.io/Personality-Aware-Non-verbal-Behavior-Generation/" target="_blank" rel="noopener">Thesis website</a></li>
<li><a class="btn btn-ghost" href="https://github.com/amramer/Personality-Aware-Non-verbal-Behavior-Generation" target="_blank" rel="noopener">GitHub repository</a></li>
<li><a class="btn btn-ghost" href="https://github.com/amramer/Personality-Aware-Non-verbal-Behavior-Generation/blob/main/docs/Thesis_final_doc.pdf" target="_blank" rel="noopener">Read the full thesis (PDF)</a></li>
</ul>
</article>
<!-- Badminton -->
<article class="project" aria-labelledby="p-badminton">
<header class="project-head">
<p class="project-kind">Computer vision system</p>
<h3 id="p-badminton">Badminton-VisionAI: AI Match Analytics from Video</h3>
<p class="project-lede">Badminton-VisionAI is an <strong>end-to-end computer vision pipeline</strong> that converts badminton match videos into structured performance analytics for players and coaches. The system tracks <strong>both players and the shuttlecock</strong>, detects shot events, projects motion onto a <strong>mini-court representation</strong> and generates a downloadable <strong>coach-style performance report</strong>.</p>
</header>
<div class="media-pair">
<figure class="media media-dark">
<video class="autoplay" muted loop playsinline preload="none" poster="assets/img/badminton-tracking-poster.webp" width="1280" height="542" aria-label="Badminton tracking: players, shuttle, shots and court mapping">
<source src="assets/media/badminton-tracking.mp4" type="video/mp4">
</video>
<figcaption><strong>Tracking and on-court intelligence.</strong> Player tracking, shuttle tracking, shot detection, court mapping.</figcaption>
</figure>
<figure class="media media-dark">
<video class="autoplay" muted loop playsinline preload="none" poster="assets/img/badminton-dashboard-poster.webp" width="1280" height="542" aria-label="Streamlit analytics dashboard for badminton tracking">
<source src="assets/media/badminton-dashboard.mp4" type="video/mp4">
</video>
<figcaption><strong>Analytics dashboard (Streamlit).</strong> Player statistics, heatmaps, shot analysis, coach-ready report.</figcaption>
</figure>
</div>
<div class="project-body">
<div>
<h4>Technical challenges</h4>
<p>Badminton tracking presents several challenges for computer vision systems:</p>
<ul>
<li><strong>Extremely small and fast shuttlecock</strong></li>
<li><strong>Frequent player occlusions</strong></li>
<li><strong>Rapid direction changes that break naive trackers</strong></li>
</ul>
<p>This project combines <strong>object detection, multi-object tracking, geometric projection and temporal analytics</strong> to produce <strong>coach-readable match insights</strong>, not just raw detections.</p>
</div>
<div>
<h4>Key features</h4>
<ul>
<li>Stable <strong>multi-object player tracking</strong></li>
<li><strong>Shuttlecock trajectory tracking</strong> with speed estimation</li>
<li><strong>Shot type classification</strong> with temporal stabilization</li>
<li><strong>Mini-court projection</strong> using homography</li>
<li><strong>Interactive analytics dashboard</strong> for match analysis</li>
<li><strong>Automated coach report</strong> with movement and shot statistics</li>
</ul>
<h4>Tech stack</h4>
<ul class="chips">
<li>Python</li><li>OpenCV</li><li>YOLO</li><li>Multi-object tracking</li><li>Court homography</li><li>Trajectory projection</li><li>Shot detection</li><li>Temporal label stabilization</li><li>Streamlit</li><li>Plotly</li><li>ReportLab PDF export</li>
</ul>
</div>
</div>
<ul class="links">
<li><a class="btn" href="https://www.youtube.com/watch?v=dAe9e_1AGuA" target="_blank" rel="noopener">Pipeline demo video: player tracking and shot analysis</a></li>
<li><a class="btn" href="https://www.youtube.com/watch?v=Bc2JLifgjzI" target="_blank" rel="noopener">Dashboard demo video: match insights and heatmaps</a></li>
<li><a class="btn btn-ghost" href="https://github.com/amramer/Badminton-visionAI" target="_blank" rel="noopener">GitHub repository</a></li>
<li class="live">
<a class="btn btn-ghost" href="https://badminton-visionai-web.onrender.com/" target="_blank" rel="noopener">Live demo (Streamlit app)</a>
<span class="note">Runs on a free server that sleeps when idle, so it can take about a minute to wake up.</span>
</li>
</ul>
<p class="closing">This project reflects my interest in <strong>real-world computer vision systems that transform model outputs into actionable insights.</strong></p>
</article>
<!-- AV sensor DataOps -->
<article class="project" aria-labelledby="p-av">
<header class="project-head">
<p class="project-kind">Data pipelines and MLOps</p>
<h3 id="p-av">AV Sensor DataOps: From Raw Sensor Logs to a Monitored 3D Detection API</h3>
<p class="project-lede">Automated pipeline that turns raw <strong>autonomous-vehicle sensor logs</strong> (nuScenes camera and LiDAR) into <strong>versioned, quality-checked, ML-ready datasets</strong>, then trains a camera detector, <strong>fuses it with LiDAR for 3D object detection</strong>, evaluates it nuScenes-style and serves it as a <strong>monitored API</strong>.</p>
</header>
<figure class="media pipeline" aria-label="Pipeline: raw sensor logs to bronze, silver, quality gate, gold, ML-ready, train, evaluate, promotion gate, ONNX and API">
<ol class="flow">
<li><span>Raw logs</span><small>DVC</small></li>
<li><span>Bronze</span><small>Parquet</small></li>
<li><span>Silver</span><small>synced, features</small></li>
<li class="gate"><span>Quality gate</span><small>schema, sensor, label checks</small></li>
<li><span>Gold</span><small>tags, splits, labels</small></li>
<li><span>Train</span><small>YOLOv8, MLflow</small></li>
<li><span>Evaluate</span><small>per-scenario slices</small></li>
<li class="gate"><span>Promotion gate</span><small>regression checks</small></li>
<li><span>Serve</span><small>ONNX, FastAPI</small></li>
</ol>
<figcaption>Every stage is a DVC stage with hashed inputs and outputs. Airflow triggers training when a new dataset version lands, and Prometheus and Grafana monitor data and model health.</figcaption>
</figure>
<div class="project-body">
<div>
<h4>What it does</h4>
<ul>
<li><strong>Versioning and lineage:</strong> every dataset and model is reproducible from a Git tag with DVC; MLflow runs carry the commit and dataset hash</li>
<li><strong>Data-quality gate:</strong> Pandera schemas plus sensor checks (sync offsets, dropped frames, corrupt files, degraded LiDAR, suspect calibration) and label checks; errors fail the pipeline</li>
<li><strong>Orchestration:</strong> Airflow with data-aware scheduling, so a new dataset version triggers training, evaluation and promotion</li>
<li><strong>Camera + LiDAR fusion:</strong> 2D detections lifted to 3D with LiDAR frustums, evaluated with nuScenes-style mAP, ATE, ASE, AOE and NDS-lite</li>
<li><strong>Serving and monitoring:</strong> ONNX Runtime and FastAPI with Prometheus metrics, Grafana dashboards and a latency benchmark for CPU, GPU and Jetson targets</li>
</ul>
</div>
<div>
<h4>CI/CD</h4>
<p>GitHub Actions runs lint and tests, the full DVC pipeline on synthetic data, the promotion gate, a reproducibility check and an API smoke test on every push, then publishes the pipeline image. A scheduled workflow re-runs the data stages and opens a pull request, so dataset releases are reviewed like code.</p>
<h4>Tech stack</h4>
<ul class="chips">
<li>DVC</li><li>Airflow</li><li>PySpark</li><li>pandas / PyArrow</li><li>Pandera</li><li>MLflow</li><li>YOLOv8</li><li>ONNX Runtime</li><li>FastAPI</li><li>Prometheus</li><li>Grafana</li><li>Docker</li><li>Kubernetes</li><li>GitHub Actions</li>
</ul>
</div>
</div>
<ul class="links">
<li><a class="btn btn-ghost" href="https://github.com/amramer/av-sensor-dataops" target="_blank" rel="noopener">GitHub repository</a></li>
</ul>
</article>
</section>
<!-- ============ OTHER PROJECTS ============ -->
<section class="wrap" aria-labelledby="h-other">
<h2 id="h-other">Other projects</h2>
<div class="cards">
<article class="card" aria-labelledby="c-driving">
<figure class="media media-dark">
<video class="autoplay" muted loop playsinline preload="none" poster="assets/img/driving-segmentation-poster.webp" width="640" height="360" aria-label="Autonomous driving segmentation results">
<source src="assets/media/driving-segmentation.mp4" type="video/mp4">
</video>
</figure>
<p class="project-kind">Computer vision, autonomous driving</p>
<h3 id="c-driving">Semantic Segmentation for Autonomous Vehicles</h3>
<p>End-to-end <strong>semantic segmentation</strong> of urban street scenes for autonomous driving perception.</p>
<ul>
<li><strong>Dataset:</strong> BDD100K</li>
<li><strong>Task:</strong> multi-class semantic segmentation</li>
<li><strong>Classes:</strong> road, traffic light, traffic sign, vehicle, person, bicycle, background</li>
<li><strong>Evaluation:</strong> mean Intersection over Union (mIoU)</li>
<li><strong>Results:</strong> mIoU ≈ 0.45, with strong performance on dominant classes (road ≈ 0.88, vehicle ≈ 0.78)</li>
<li><strong>Optimization:</strong> systematic hyperparameter optimization via experiment sweeps</li>
<li><strong>Pipeline:</strong> data preparation → training → evaluation → deployment and visualization</li>
</ul>
<ul class="chips"><li>PyTorch</li><li>fastai</li><li>Semantic segmentation</li><li>Hyperparameter optimization</li><li>Experiment tracking</li></ul>
<a class="card-link" href="https://github.com/amramer/Semantic-Segmentation-Model-for-Autonomous-Vehicles-An-End-to-End-ML-Workflow" target="_blank" rel="noopener">GitHub repository</a>
</article>
<article class="card" aria-labelledby="c-brain">
<figure class="media">
<video class="autoplay" muted loop playsinline preload="none" poster="assets/img/brain-tumor-segmentation-poster.webp" width="762" height="408" aria-label="Brain tumor segmentation on MRI">
<source src="assets/media/brain-tumor-segmentation.mp4" type="video/mp4">
</video>
</figure>
<p class="project-kind">Medical imaging</p>
<h3 id="c-brain">3D Brain Tumor Segmentation (MRI)</h3>
<p>Multi-label <strong>3D semantic segmentation</strong> of glioma sub-regions from volumetric MRI scans.</p>
<ul>
<li><strong>Dataset:</strong> Medical Segmentation Decathlon / BraTS (multi-modal MRI)</li>
<li><strong>Model:</strong> 3D SegResNet (MONAI)</li>
<li><strong>Input modalities:</strong> FLAIR, T1, T1Gd, T2</li>
<li><strong>Target structures:</strong> whole tumor (WT), tumor core (TC), enhancing tumor (ET)</li>
<li><strong>Training and evaluation:</strong> Dice loss and mean Dice</li>
<li><strong>Results:</strong> mean Dice = 0.78 on validation (WT 0.90, TC 0.82, ET 0.59)</li>
<li><strong>Pipeline:</strong> preprocessing → training → inference → deployment and visualization</li>
</ul>
<ul class="chips"><li>PyTorch</li><li>MONAI</li><li>3D SegResNet</li><li>Multi-modal MRI</li><li>3D medical image transforms</li><li>Sliding-window inference</li><li>Weights & Biases</li></ul>
<a class="card-link" href="https://github.com/amramer/brain-tumor-segmentation-3D-DeepLearning" target="_blank" rel="noopener">GitHub repository</a>
</article>
<article class="card" aria-labelledby="c-vlm">
<figure class="media media-dark">
<video class="autoplay" muted loop playsinline preload="none" poster="assets/img/realtime-captioning-poster.webp" width="1280" height="674" aria-label="Realtime vision captioning app">
<source src="assets/media/realtime-captioning.mp4" type="video/mp4">
</video>
</figure>
<p class="project-kind">Vision-language models</p>
<h3 id="c-vlm">Realtime Vision Captioning</h3>
<p>Exploratory project focused on applying <strong>vision-language models in realtime webcam-based settings</strong>.</p>
<ul>
<li>Realtime image captioning and visual question answering using <strong>BLIP</strong> via <strong>Hugging Face Transformers</strong></li>
<li>Image classification using <strong>ResNet-50</strong> pretrained on <strong>ImageNet</strong></li>
<li>Realtime <strong>Gradio-based webcam application</strong> for live image captioning and classification with adjustable inference parameters</li>
</ul>
<p>The project is implemented as a set of Jupyter notebooks, progressing from offline image understanding to <strong>a realtime vision application</strong>.</p>
<ul class="chips"><li>Python</li><li>PyTorch</li><li>Torchvision</li><li>Hugging Face Transformers</li><li>Gradio</li><li>PIL</li><li>NumPy</li></ul>
<a class="card-link" href="https://github.com/amramer/realtime-vision-captioning" target="_blank" rel="noopener">GitHub repository</a>
</article>
<article class="card" aria-labelledby="c-agent">
<figure class="media media-dark">
<img src="assets/img/conversational-agent.webp" alt="AI conversational agent web interface with an avatar" width="1600" height="764" loading="lazy" decoding="async">
</figure>
<p class="project-kind">Conversational AI <span class="status">Ongoing</span></p>
<h3 id="c-agent">AI Conversational Agent with Voice, Avatar and Web Interface</h3>
<p>Conversational AI system enabling real-time voice and text interaction through a custom web-based interface.</p>
<ul>
<li>Speech-to-text and text-to-speech for natural dialogue</li>
<li>LLM-powered conversational reasoning</li>
<li>Human-like conversational avatar integrated into the UI</li>
<li>Modular frontend-backend design for extensibility</li>
</ul>
<ul class="chips"><li>Python</li><li>OpenAI GPT</li><li>Speech-to-text</li><li>Text-to-speech</li><li>HTML</li><li>CSS</li><li>JavaScript</li><li>Bootstrap</li><li>jQuery</li></ul>
</article>
</div>
</section>
<!-- ============ SKILLS ============ -->
<section class="wrap" id="skills" aria-labelledby="h-skills">
<h2 id="h-skills">Technical profile</h2>
<dl class="skills">
<div><dt>Computer vision tasks</dt><dd>Object detection, segmentation, tracking, 3D object detection, camera and LiDAR sensor fusion, video motion analysis, digital avatar generation</dd></div>
<div><dt>Models and frameworks</dt><dd>PyTorch, TensorFlow, OpenCV, YOLO, SAM, CNNs, Vision Transformers (ViT), TrackNet, Supervision</dd></div>
<div><dt>Training and evaluation</dt><dd>Transfer learning, fine-tuning, loss design, metric selection, hyperparameter optimization, multi-GPU training (CUDA)</dd></div>
<div><dt>Data pipelines and quality</dt><dd>Bronze, silver and gold data layers, DVC dataset versioning and lineage, Airflow orchestration, PySpark, pandas / PyArrow, Parquet, Pandera schema and data-quality checks</dd></div>
<div><dt>Data and experimentation</dt><dd>Dataset preparation, data augmentation, efficient data loading, MLflow, Weights & Biases, TensorBoard, ablation studies</dd></div>
<div><dt>Deployment and inference</dt><dd>Docker, Kubernetes, FastAPI, ONNX Runtime, GPU inference pipelines, batch and real-time inference, AWS (EC2 GPU), Streamlit (demos)</dd></div>
<div><dt>CI/CD and monitoring</dt><dd>GitHub Actions, automated testing and reproducibility checks, model promotion gates, Prometheus, Grafana</dd></div>
<div><dt>Compute and infrastructure</dt><dd>SLURM (HPC), GPU job scheduling, multi-node training, large-scale GPU experiments</dd></div>
<div><dt>Software engineering foundations</dt><dd>Python, C++, object-oriented design, version control (Git), debugging, logging, unit testing</dd></div>
</dl>
</section>
<!-- ============ BEYOND WORK ============ -->
<section class="wrap" id="beyond" aria-labelledby="h-beyond">
<h2 id="h-beyond">Beyond work</h2>
<p class="section-lede">Outside of my professional work, I enjoy aerial photography and videography using drones. This section highlights selected visual projects created in my free time.</p>
<div class="vimeo-grid">
<div class="vimeo"><iframe src="https://player.vimeo.com/video/1152190588?dnt=1" title="Drone video 1" loading="lazy" allow="fullscreen; picture-in-picture" allowfullscreen></iframe></div>
<div class="vimeo"><iframe src="https://player.vimeo.com/video/1152190476?dnt=1" title="Drone video 2" loading="lazy" allow="fullscreen; picture-in-picture" allowfullscreen></iframe></div>
<div class="vimeo"><iframe src="https://player.vimeo.com/video/1152190673?dnt=1" title="Drone video 3" loading="lazy" allow="fullscreen; picture-in-picture" allowfullscreen></iframe></div>
</div>
<p><a href="https://vimeo.com/amramer252805454" target="_blank" rel="noopener">View more drone videos on Vimeo</a></p>
</section>
<!-- ============ CONTACT ============ -->
<section class="wrap contact" id="contact" aria-labelledby="h-contact">
<h2 id="h-contact">Contact</h2>
<p class="section-lede">If you're hiring or would like to collaborate, I'd be glad to hear from you.</p>
<ul class="contact-list">
<li><span>Email</span><a href="mailto:amribrahim.amer@gmail.com">amribrahim.amer@gmail.com</a></li>
<li><span>LinkedIn</span><a href="https://www.linkedin.com/in/amr-amer-2023-cs/" target="_blank" rel="noopener">linkedin.com/in/amr-amer-2023-cs</a></li>
<li><span>GitHub</span><a href="https://github.com/amramer" target="_blank" rel="noopener">github.com/amramer</a></li>
<li><span>Phone</span><a href="tel:+491626667294">+49 162 6667294</a></li>
</ul>
</section>
</main>
<footer class="foot wrap">
<p>© 2026 Amr Amer</p>
<p><a href="assets/Resume_Amr_Amer.pdf" download="Amr_Amer_CV_ML_Engineer.pdf">Download CV (PDF)</a></p>
</footer>
<script src="assets/js/main.js" defer></script>
</body>
</html>