AI/ML Engineer building production-ready ML and Generative AI systems across modeling, serving, evaluation, and monitoring.
From validated data pipelines → model evaluation → deployed APIs → decision-ready dashboards.
| Area | What you can expect |
|---|---|
| Production ML systems | Leak-safe evaluation, calibrated models, threshold policies, reproducible artifacts, and decision-ready outputs |
| Generative AI workflows | RAG evaluation, retrieval attribution, tool-calling agents, structured outputs, and quality-aware rollout review |
| APIs & serving | Dockerized FastAPI services, strict request/response schemas, versioned artifacts, and CI-friendly delivery |
| Monitoring & decision support | Telemetry, observability, drift and risk review, latency and cost trade-offs, triage logic, and operational dashboards |
| Applied NLP & CV | Text classification, semantic search, threshold tuning, explainability, and practical computer vision apps |
| Project | Focus | Links |
|---|---|---|
| Fraud Detection Dashboard | Streamlit analytics UI + FastAPI inference + threshold tuning + cost-aware fraud review | Repo |
| LLM Production Telemetry | Multi-table telemetry → routing policies, drift review, triage thresholds, and decision artifacts | Repo · Data |
| RAG QA Logs & Corpus | Retrieval attribution, failure taxonomy, KPI baselines, trade-off analysis, and rollout gating | Repo · Data |
| Road Accident Risk | Residual-boosted ensemble, stable OOF evaluation, interpretable risk features, and calibrated safety scoring | Kaggle |
| EV Charging Analytics | Geospatial analytics, fast-DC allocation optimizer, market slices, and scenario-based network planning | Repo |
| Cancer Risk Analysis | Clean tabular data, validation, leakage-aware benchmarking, and interpretable risk modeling | Repo · Data |
| Project | Focus | Links |
|---|---|---|
| Advanced ML Sentiment Lab | Streamlit + Plotly sentiment analysis lab with TF-IDF, ROC/PR evaluation, threshold tuning, error analysis, and live prediction | Repo |
| Text Sentiment Analysis | IMDB sentiment pipeline with calibrated TF-IDF baselines, threshold tuning, explainability, and BiLSTM baseline | Repo |
| SMS Spam Detection | Dual TF-IDF pipeline with calibrated Linear SVM, nested CV, threshold tuning, explainability, and robustness checks | Repo |
| Old Photo Restorer | Practical computer vision Gradio app for old-photo restoration with GFPGAN, optional upscaling, and batch ZIP export | Repo |
| Dataset | What it’s for | Links |
|---|---|---|
| Cancer Risk Factors | Clean health, lifestyle, environmental, and genetic features for EDA and risk modeling | Kaggle |
| Global EV Infrastructure | Standardized EV charging infrastructure data for geospatial analytics, planning, and network modeling | Kaggle |
| YouTube Shorts & TikTok Trends 2025 | Short-form content analytics, trend exploration, and virality analysis | Kaggle |
| Digital Lifestyle & Mental Wellness | Behavioral signals for wellbeing analytics and predictive workflows | Kaggle |
| Category | Tools |
|---|---|
| Languages & Core |       |
| Data & Analytics |      |
| ML / DL |      |
| NLP / CV / LLM |           |
| Apps & Dashboards |        |
| APIs & Serving |          |
| Monitoring & Quality |        |
- 🚀 Production ML & Generative AI systems: FastAPI services, Dockerized delivery, evaluation-first workflows, and decision-ready outputs
- 🧠 RAG & LLM reliability: retrieval attribution, grounded outputs, guardrails, and regression-friendly review
- 🗂️ Datasets & reproducible assets: validated schemas, clear documentation, reusable notebooks, and ML-ready data products
- 📊 Decision-ready dashboards: interactive analytics, threshold tuning, monitoring, and operational insights
Best contact: LinkedIn
If you find the work useful, a ⭐ helps more people discover it.