I'm a final-year B.Tech Computer Science student at Amity University, Noida (8.86 CGPA). I build backend and applied AI systems for pipeline monitoring, code exploration, and document retrieval. As an AI Engineering Intern at Marksman Technologies, I helped ship a lead qualification tool used by the sales team.

I'm interested in software engineering roles focused on backend systems, data infrastructure, and applied AI.

PipeGuard monitors an Airflow and dbt workload, records execution attempts and data quality results, and traces which datasets a run touches. It preserves earlier failures after a successful retry. Operators can review incident evidence, request an AI investigation, and separately approve a scoped recovery action.

Stack · Python, FastAPI, PostgreSQL, Airflow, dbt, OpenLineage, Next.js

Verification · 599 automated tests and both hosted CI jobs passed at the November 2026 checkpoint.

• CritHop · Answers multi-hop questions by traversing a passage graph and checking evidence relevance and support along the way. Phase 2 uses a trained reranker to score passage relevance.
• Codebase Oracle · Parses repositories into a symbol and dependency graph. Combines 2D and 3D exploration with GraphRAG answers that cite source files and line ranges.
• AirForce · Models indoor air quality in a 3D floor-plan builder. Uses browser-based simulation to show how airflow and pollution change across rooms.
• reranker-slm · Fine-tunes Qwen2.5-0.5B with QLoRA on MS MARCO passage pairs for relevance scoring. The adapter powers CritHop's trained relevance check.
• API Docs RAG · Searches 1,220 GitHub REST operations from the OpenAPI spec using dense and keyword retrieval. Returns answers with source links and abstains when evidence is insufficient.

• BigQuery Release Radar · Search and filter BigQuery release notes in a Flask dashboard, with caching and CSV export.
• Brand Builder · Generate product advertising mockups for billboard, newspaper, and social formats.
• The Tiebreaker · Compare options using weighted criteria, pros and cons, and SWOT analysis.
• Car Sales Prediction · Compare linear regression and tuned random forests on a customer purchasing capacity dataset.
• DSA solutions · Topic-organized C++ solutions and a separate LeetCode practice log.

In CritHop's documented evaluation, the Phase 2 reranker improved exact match (EM) over the reported HopRAG baseline on all three datasets.

Read the CritHop benchmark results

API Docs RAG retrieved the expected operation in its top three results on 20 of 29 evaluation queries (68.97% Hit@3).

Read the API Docs RAG evaluation

Snapshot from the public profile on 8 November 2026. Each classified repository contributes one count to its primary language. View live contribution history

• Marksman Technologies · AI Engineering Intern. Helped build a hybrid lead qualification engine and sales intelligence tools.
• Amity University, Noida · B.Tech in Computer Science and Engineering, 8.86 CGPA.