|
CS undergraduate at IIIT Naya Raipur with hands-on experience designing scalable backend systems, distributed software architectures, and production-grade AI applications. Open-source contributor to PyTorch and vLLM / LLM Compressor, with a maintainer-merged PR and a maintainer-approved PR in core ML infrastructure. Selected for Amazon ML Summer School 2026. What I bring:
β Β Scalable, distributed data & ML pipelines (Databricks, PySpark, Delta Lake)
|
|
Scikit-learn Β Β·Β MLflow Β Β·Β NLP Β Β·Β RAG Β Β·Β Time-Series Analysis Β Β·Β Databricks Β Β·Β PySpark Β Β·Β Delta Lake Β Β·Β Docker Compose Β Β·Β JWT / RBAC Β Β·Β FAISS Β Β·Β Distributed Systems Β Β·Β System Design
| Project | Contribution | Status |
|---|---|---|
| PyTorch (102kβ , ~21M PyPI dl/wk) | Fixed return typing of torch.fx.graph.CodeGen.process_inputs using TypeVarTuple/Unpack to preserve variadic argument types (zero runtime change) β approved by maintainer @Skylion007 |
π‘ Open (PR #188287) |
| vLLM β LLM Compressor (3.5kβ ) | Modernized legacy Optional[Dict[...]] typing to native dict[...] | None syntax in the sparsification utility and removed unused imports |
β Merged (PR #2870) |
| vLLM β LLM Compressor (3.5kβ ) | Built a post-calibration guard raising ValueError on missing/non-finite/non-positive KV-cache scales; refactored into a reusable validate_module_calibration pattern per maintainer @kylesayrs β iterated across 9 review commits |
π‘ Open (PR #2887) |
|
Healthcare ML Pipeline β Databricks Lakehouse End-to-end cardiovascular risk prediction from PPG signals using Databricks Lakehouse (BronzeβSilverβGold Delta Lake), PySpark, and MLflow, with reliability-adjusted late-fusion across PPG, vitals, OCR reports, and symptom history. 79.49% accuracy / 0.796 F1 on subject-wise GroupKFold (15 subjects, 6,856 windows) |
Multi-Agent AI Research Platform Production-oriented platform orchestrating 8 specialized agents (RAG, Math, Code, Citation, GitHub, Podcast) across 36 Flask routes with async job queues, bounded backpressure, and progress-polling endpoints.
Runs without external API keys β lexical + optional FAISS/sentence-transformer semantic search. |
AI-Powered Applicant Tracking System Production-ready ATS with JWT authentication, role-based access control (ADMIN / RECRUITER / CANDIDATE), 37 REST API routes, and React dashboards with request-traced audit logging and TTL caching. 86% accuracy / 84% F1 across a 41-check live smoke suite |
|
Hospital Management System Role-based hospital platform for Admin, Doctor, Receptionist & Pharmacist with billing, pharmacy, appointment scheduling, patient records, and BCrypt-secured authentication.
|
Civic Issue Resolution Platform Location-based complaint tracking with AI-assisted categorization, duplicate detection, severity handling, and geospatial task assignment across citizen, official, and worker roles.
|
Agricultural Marketplace Platform Digital farm-to-buyer marketplace with smart crop matching, location proximity scoring, bidding workflows, advisory notifications, and JWT-secured role-based access.
|
| Platform | Standing | Core Topics |
|---|---|---|
| π‘ Β LeetCode / Codeforces (combined) | 400+ Problems Solved | Arrays Β· DP Β· Trees Β· Graphs Β· Sliding Window |
| π΅ Β Codeforces | Pupil Β· Max Rating: 1263 | Greedy Β· BFS / DFS Β· Binary Search Β· Backtracking |
| π Achievement | Detail |
|---|---|
| π Amazon ML Summer School | Selected for 2026 β competitive program for top undergraduates across India |
| π Open-Source | Merged PR in vLLM / LLM Compressor; approved PR in PyTorch (102kβ ) |
| π Codeforces Round 1104 (Div. 1+2) | Ranked 2,458 / 24,325 |
| π Codeforces Round 1054 (Div. 3) | Ranked 7,457 / 63,421 |
| β‘ Codeforces | Pupil, Max Rating 1263; 400+ DS&A problems solved |
| π― JEE Main 2024 | 98.25 percentile β Top 1.75% of 1.2M candidates nationwide |
| π Model Quality | CardioTwin AI β 79.49% accuracy / 0.796 F1 on subject-wise GroupKFold |
- Databricks Accredited Generative AI Fundamentals β Jun 2026
- Data Science & Analytics β HP LIFE / HP Foundation β Jun 2026
π§ Distributed Systems Β· System Design Β· Scalable Backend Engineering
βοΈ MLOps Β· Databricks Β· PySpark Β· Delta Lake Β· Experiment Tracking
π RAG Systems Β· LLM Fine-tuning Β· Prompt Engineering
π Deepening contributions to PyTorch & vLLM / LLM Compressor
π― Open-Source AI / ML Infrastructure Contributions
Seeking internship opportunities in AI/ML, backend/distributed systems, and software engineering.
Open-source contributor to PyTorch and vLLM β I bring production-grade ownership from data pipelines to deployed, documented systems.