Changelog

v2026.06: AI & ML Associate, Impact Solutions Lab

Converted from intern to a full-time AI & ML Associate at Impact Solutions Lab (an AAM Foundation initiative) on 1 June 2026. What I'm building now:

A Text-to-SQL survey-intelligence pipeline that lets nonprofit staff query 3,200+ community survey responses in plain English. Groq (Llama 3.3 70B) with schema-aware prompting, read-only SQL validation with auto-repair, and a narration layer that turns query results into concrete program recommendations. The messy multilingual data lives in a hybrid SQLite schema (a shared core table plus an EAV overflow table) so it never needs a migration when new surveys arrive.
A per-student learning chatbot grounded in each student's own class recordings. A WhisperX Hindi-and-English transcription pipeline with multi-layer hallucination filters feeds a pgvector store of 1,013 chunks across 9 classes; every student gets a RAG bot scoped to what they said, missed, and were taught, with strict per-student isolation. Shipped behind a security audit and a passing 367-test suite.
Rebuilding the entrance-exam portal into an all-purpose internal assessment platform: teacher content upload (Word to JSON), student and subject assignment, and role-based access.

v2026.05: Graduated

B.Tech in Computer Science Engineering (AI & ML) from Jain University. Shipped along the way: one patent application, one published paper, and one adversarial Wumpus.

v2025.10: AI & ML Intern, Impact Solutions Lab

Oct 2025 to May 2026. Built the internal tooling that keeps a growing education program running:

A Drive-to-YouTube pipeline that scans Drive folders, extracts class recordings from ZIP archives, uploads them to YouTube, logs the links to Google Sheets, and runs itself on GitHub Actions.
A Zoom attendance tool that pulls participant data and turns it into per-student attendance reports.
The Adira Academy entrance-exam portal, used by 300+ students for application, assessment, and selection.
Early work on the AI for Social Impact DataHub, exploring community survey data and laying the groundwork for a production Text-to-SQL pipeline.

v2025.02: Machine Learning Intern, SixPhrase

Feb to May 2025. End-to-end classical ML: preprocessing, PCA/LDA that cut training time ~25%, supervised models (logistic regression, decision trees, SVM, KNN) up to 87% accuracy, and L1/L2/dropout regularization.

v2024.04: Data Analysis Intern, Miles Education

Apr to May 2024. A hands-on analytics rotation, start to finish:

SQL cleaning and modelling that cut redundancy ~30%.
Exploratory analysis in Python (Pandas, Matplotlib, Seaborn) on real datasets.
A Power BI hotel-bookings dashboard surfacing customer-behaviour and seasonal trends.
A group capstone report and weekly stakeholder presentations.

v2022: Volunteer Core Team Member, GDSC

2022 to 2023. Program ops for the campus Google Developer Student Clubs chapter: organised and promoted events, coordinated the core team, and kept communication flowing so event days ran smoothly.