ML/Full-Stack Engineer | Deep Learning | AI Systems | Production Code

Building production ML systems & scalable backends. 5/5 Deep Learning courses completed. 100+ DSA problems solved. Google Gemini Campus Ambassador.

Languages: Python | TypeScript | Java | C
ML/AI: TensorFlow | XGBoost | Scikit-learn | Gemini API | Prompt Engineering
Backend: Flask | FastAPI | Node.js | Express.js | PostgreSQL | MongoDB
Frontend: React.js | Next.js
DevOps: Docker | GitHub Actions | Git | Pytest | Vitest

** ML/DL Vocational Trainee ** — IIIT Naya Raipur (June – July 2025)

• Built supervised ML system on MIMIC-III (48K samples, 31 subjects) for cuffless BP prediction
• Achieved MAE 6.84/3.42 mmHg — approaching clinical benchmarks via subject-independent validation
• Optimized preprocessing pipeline with vectorized ops: 40% runtime reduction
• Demonstrates: Medical-grade ML, signal processing, production pipelines

Cuffless Blood Pressure Estimation from PPG Signals (In Progress) | TensorFlow, CNN-BiLSTM, Python

• Research project on MIMIC-III dataset (48K samples, 31 subjects) for non-invasive clinical monitoring
• Achieved MAE 6.84/3.42 mmHg (Systolic/Diastolic) — approaching clinical benchmarks via subject-independent validation
• 3-channel CWT scalograms (PPG, VPG, APG) + temporal attention mechanisms
• Demonstrates: Medical-grade ML, signal processing, research-quality evaluation metrics

Heart Rate Estimation via 1D-CNN on PPG Signals (Deployed) | TensorFlow, Flask, Docker, Python

• Trained on BIDMC ICU dataset (53 subjects, 8-second windows) with patient-held-out validation
• Tripled training data via domain-specific augmentation (noise injection, amplitude scaling, temporal shifting)
• Achieved 3.5 BPM MAE — Dockerized Flask REST API with Pytest automation for reproducible inference
• Demonstrates: Production deployment, signal processing, data augmentation strategy

ReadyCheck AI | Next.js, TypeScript, Gemini API, PostgreSQL

• Multi-tenant assessment platform with Row Level Security (RLS) at database layer
• Fault-tolerant JSON parser for truncated LLM outputs — backend reliability by design
• 261 passing tests across 14 test files | Features: Dynamic question generation, timer, honor code monitoring
• Demonstrates: Full-stack + GenAI, HCI principles, comprehensive testing

Credit Card Fraud Detection Pipeline | Python, XGBoost, FastAPI, Pandas

• 99:1 class imbalance solved via sequential resampling (no temporal leakage)
• O(N) rolling transaction velocity windows — algorithmic optimization for fraud pattern detection
• In-memory model caching + threshold tuning: 90.22% fraud recall
• Demonstrates: Real-world ML, interpretability (SHAP), production pipeline design

LocalConnect — Service Marketplace | Node.js, Express.js, MongoDB

• 5-state FSM for booking lifecycle (PENDING → CONFIRMED/COMPLETED/CANCELLED)
• Prevented double-booking with MongoDB compound indexes (interval-overlap queries)
• Incremental rating computation on write (vs. expensive aggregation on read)
• Demonstrates: Backend system design, database optimization, state machine architecture

Neural Networks from Scratch | Python, NumPy

• Implemented backpropagation & gradient descent using only matrix calculus
• Breast Cancer Classifier: 93.86% accuracy, 88.37% malignant recall
• MNIST Softmax on 70,000 samples
• Demonstrates: Deep understanding of ML fundamentals, mathematical foundations

Deep Yet Simple — Technical Blog Platform | Next.js, React, Tailwind CSS

• Modern blogging platform with server-side rendering & SEO optimization
• Demonstrates: Full-stack development, performance optimization, modern web technologies

✓ Google Gemini Campus Ambassador — Led college-wide AI adoption workshops
✓ Deep Learning Specialization (5/5 courses) — deeplearning.ai / Andrew Ng
✓ Lenovo LEAP NextGen Scholar — Full-Stack Web Development (MERN)
✓ 100+ DSA Problems — Trees, Graphs, DP, Sliding Window (complexity-optimized)

B.Tech Information Technology — Government Engineering College Jagdalpur (CSVTU)
Expected Graduation: 2027 | Current: Semester 6 (April 2026)

Coursework: Machine Learning, Deep Learning, Neural Networks, Data Structures & Algorithms, Signal Processing, DBMS, AI

Biomedical AI | LLMs & Generative Systems | Time-Series Modeling | Production ML Systems

Roles: AI/ML Research Internship | SDE Internship | SDE Fresher | Deep Learning Engineer
Focus: Production ML systems, backend engineering, medical AI, full-stack development

GitHub: github.com/rohit-sinha-76
LinkedIn: linkedin.com/in/rohit-sinha-76
Email: work.rohit.sinha.11@gmail.com