available for opportunities
Abhinav Aditya
Data & AI Engineer
Building AI systems at scale - from LLM pipelines to MLOps.
◇ New York, NY
Experience
- Shipped a production GenAI agent over the MLS listing API, letting agents query property details in natural language; caching cut response latency 40% (5s → 3s).
- Built Metabase dashboards on MySQL to monitor housing-sales and subscription metrics, surfacing usage patterns that drove product experiments.
- Ran A/B testing and funnel analysis across 1,000+ users, driving a 19% lift in sign-ups and a 17% increase in lead conversion.
- Led a front-end revamp of the CRM in Next.js with Redux, integrating RESTful APIs and partnering with product stakeholders to define success KPIs ahead of each release.
- Established a CI/CD pipeline with GitHub Actions (78.3% test coverage), cutting build time by 57%.
- Instrumented Prometheus telemetry across Docker services for real-time anomaly detection on Radius Rooms, a WebRTC platform scaled to 500+ MAUs.
Internships
- Built and shipped web features in JavaScript, collaborating through Git-based workflows.
Senior Editor & Head of Product Development
turf FootballDelhi, India
- Led editorial and product for an Indian football platform covering news, transfers, and match analysis.
- Drove product development as Head of Product Development (Nov 2020 - Dec 2021).
- Contributed to engineering work using Git-based development workflows.
Education
- Current CGPA: 3.87 / 4.0
- Relevant coursework: MLOps, Applied Generative AI, Data Mining, Data Management for Analytics, Databases, Machine Learning.
- CGPA: 8.13 / 10.0
- Leadership: Captain of the university soccer team; co-mentor for Java at ACM MUJ.
Featured Projects
Settl - Autonomous AI Invoice Recovery
A multi-agent system that autonomously recovers unpaid invoices - a 5-stage pipeline on GCP Vertex AI Agent Builder with Gemini 3 Pro. Shipped the deterministic strategy agent and compliance gate (28 passing tests), with a source-agnostic ingestion layer and a full decision audit trail.
Workhuman HR Analytics Capstone
A 4-stage LLM document-processing pipeline over 1,000+ unstructured records - Llama-3-8B for NER + behavioral extraction, HDBSCAN clustering, and a Claude-seeded taxonomy, with SLM/LLM routing for a ~10x cost cut.
CiteConnect - Semantic Academic Search
Production MLOps pipeline (GitHub Actions, Docker, Airflow) deploying MiniLM-L6-v2 + SPECTER2 to GCP Cloud Run / Kubernetes with drift-triggered retraining, plus a RAG layer over Pinecone indexing 100K+ docs with hybrid semantic + citation-graph search.
BIBOT - Business Intelligence Bot
A LangChain multi-agent chatbot, built in a hackathon sprint, that turns user-uploaded CSVs into business-intelligence insights and visualisations.
From GitHub
Dynamic mental health website that serves as a platform where people can find a safe space to share as well as cure their mental health-related problems.
Android app that scans a family member's medical documents into per-visit folders backed up to their own Google Drive.
LeetCode practice questions and notes
An intelligent orchestration layer that cuts 90% of LLM costs by routing tasks to the correct, hyper-efficient small model paradigm.
IIT Bombay Abhyuday
Open Source
DhikshaMathanagopal/CiteConnect
Abhinav/fix dvc
Cite-Connect/CiteConnect-ModelPipeline
cron change for 20 mins will need to change that to 3 hours
carnageitself/HR-Capstone
testing change
Illapavan/humanoid-robot
Abhinav/feat/pdf reader
enrique7mc/music-posters
ci: add typecheck gate and clear the 38 pre-existing tsc errors
HKUDS/nanobot
[Security] `exec.allowPatterns` shell-chain bypass allows unintended command execution
Cite-Connect/CiteConnect-DataPipeline
Abhinav/test new changes
Stack
Beyond the Terminal
$ let's connect
Building something at the edge of AI & data?
I'm open to roles, collaborations, and good problems. My inbox is always open.
aditya.ab@northeastern.edu