# Rishabh Sharma

> Rishabh Sharma, AI Lead Engineer at Spinutech in Tampa. Author of the Engram agent-memory paper; builds OmniHarness and Archimedes; paper reviewer.

Rishabh Sharma is an AI Lead Engineer at Spinutech, based in Tampa, FL. He builds agent systems that verify their own work — long-horizon agent harnesses, autonomous research loops and measurable agent memory — and tests them with pre-registered experiments. He works on data and AI at Spinutech in Tampa, FL, reviews papers, judges hackathons, maintains the open-source archimedes.run project, and holds an MS in Data Science from Drexel University.

Contact: rishabh.sharma1103@gmail.com

## Systems
- [Engram](https://www.rishabhsharma.co/work/engram): A pre-registered test of conversational memory for AI agents: raw turns selected by a typed decision model versus engram v2, an LLM-extraction memory, at matched context.
- [Archimedes](https://www.rishabhsharma.co/work/archimedes): An open-source autonomous AI research framework that produces reproducible traces — literature map, plan, code, metrics, failures and a draft — rather than finished papers.
- [OmniHarness](https://www.rishabhsharma.co/work/omniharness): An open-source super-agent harness that orchestrates subagents, persistent memory and sandboxed execution — built on the bet that the agent should close its own loops.
- [Area O1](https://www.rishabhsharma.co/work/area-o1): A free, open-source, local-first workspace that turns your papers, code, talks, press and judging into a living, source-traced evidence file.
- [Rallion](https://www.rishabhsharma.co/work/rallion): A marketing command center that watches the market for buying signals, turns them into evidence-cited campaign opportunities a human approves, then runs the approved outreach end to end.

## Recognition
- First-author arXiv paper · 2609.34227
- Hugging Face Daily Papers · Oct 2026
- Paper reviewer
- Judge & mentor · HackMIT 2026
- Invited judge · HackPrinceton Fall 2026
- MS Data Science · Drexel Dean's Fellow
- AWS Certified Data Engineer

## Other work (all at https://www.rishabhsharma.co/work)
- [DealFlow](https://github.com/ris3abh/DealFlow): A framework for AI sales agents that hold natural, multi-turn sales conversations.
- [TAMC](https://github.com/ris3abh/TAMC): Persistent homology as a control signal for time-series test-time adaptation.
- [ResArch / Spinscribe](https://github.com/ris3abh/ResArch-core): Multi-agent content API with Cognito auth and S3 storage.
- [ResumeForgeAI](https://github.com/ris3abh/ResumeForgeAI): Tailors a resume to a job description — and to the ATS that reads it.
- [WebWiseAgent](https://github.com/ris3abh/WebWiseAgent): A personal research assistant that works the web for you.
- [AML P2P Fraud](https://github.com/ris3abh/aml-p2p-fraud-detection): Caught my own leak: AUPRC 0.988 → an honest 0.380. 26.1% SAR conversion at 2% alert capacity.
- [Home Credit Default Risk](https://github.com/ris3abh/Home-Credit-Default-Risk): Credit-risk modelling for the unbanked, shipped behind a Jenkins pipeline.
- [Sentencing Prediction](https://github.com/ris3abh/Sentencing-prediction-on-docket-data): Predicting sentencing type from court docket data, used at Drexel's Kline School of Law.
- [Wildfire Perimeters](https://github.com/ris3abh/DL-firefighters): Wildfire and perimeter detection from GeoTIFF imagery.
- [Collision Severity](https://github.com/ris3abh/Collisions_prediction): Predicting injury severity from crash data with gradient-boosted ensembles.
- [Customer Segmentation](https://github.com/ris3abh/customer_segmentation): Estimating family size from customer attributes and purchase patterns.
- [NBFC Loan Foreclosure](https://github.com/ris3abh/NBFC_Loan_forclosure_Prediction): Predicting early loan foreclosure for a non-bank lender.
- [GoodReads Pipeline](https://github.com/ris3abh/The_GoodReads_Pipeline): Data pipeline for GoodReads API data into a Redshift warehouse.
- [Airflow + EMR ETL](https://github.com/ris3abh/ETL-pipeline-using-Airflow-and-AWS-EMR): Airflow-orchestrated Spark jobs on EMR, from S3 back to S3.
- [Covid-19 Forecasting](https://github.com/ris3abh/DSCI521Covid-19DashBoard): Forecasting cases and deaths in Mexico, with a dashboard.

## Profiles
- [GitHub](https://github.com/ris3abh)
- [archimedes.run](https://www.archimedes.run)
- [Hugging Face](https://huggingface.co/ris3abh-11)
- [arXiv](https://arxiv.org/abs/2609.34227)
- [LinkedIn](https://www.linkedin.com/in/rishabh-sharma-06b4a3188/)