Deven Shah
AI Researcher & Experienced Software Engineer
About Me
Overview
Software engineer and AI researcher focused on turning research into production. I advance medical imaging and deepfake detection at Boston University, where I am pursuing an M.S. in Computer Science. My background includes leading product vision at Suno Analytics, building member-facing applications at Patelco, and ensuring safe and secure data operations at NetApp.
Skills
Experience
Professional Journey
Education
Technical Architecture Analyst
Graduate Research Assistant
Conducting computer vision and AI research across two labs at Boston University, across biotechnology and criminal justice.
- Digital Forensics Research Lab (Mar 2026 – Present): Researching deepfake generation and detection landscape alongside faculty and peers across AI and criminal justice, building frameworks that hold up against real-world media integrity threats.
- Artificial Intelligence and Computer Vision Lab (Oct 2025 – Present): Researching automated osteoarthritis diagnosis alongside faculty, medical professionals, and peers across computer vision and clinical radiology, building pipelines that replace hours of manual MRI review.
Graduate Teaching Assistant
Teaching assistant for CS 566: Analysis of Algorithms.
Build Student Consultant
Built AI-powered image retrieval and drone path planning systems in Python, leveraging PyTorch, FAISS, Qdrant, and Docker for scalable semantic search and dataset capture.
Co-Founder & CTO
Co-founded Suno Analytics and led technical vision from prototype to production, delivering an AI-powered enterprise resource planning (ERP) platform for ecommerce companies.
- Spearheaded a globally distributed Agile engineering team across backend, frontend, and ML, driving rapid feature development and client issue resolution to maintain SLA standards.
- Architected a full-stack web platform (React + TypeScript frontend, PostgreSQL backend) with OAuth authentication and WebSocket-based connections for real-time updates.
- Engineering a forecasting system to generate purchase orders with ~80% confidence and maintain reliable stock levels through DeepAR and PyTorch, trained on historical sales and seasonal patterns.
- Launched AI-powered analytics features that increased client retention and platform adoption.
Application Developer
Shipped member-facing and internal banking applications to 500K+ users, working with product managers and stakeholders to convert requirements into compliant production systems.
- Eliminated required in-branch visits by architecting a virtual fraud request system (Visualforce frontend + Apex backend) enabling members to submit and track fraud claims digitally.
- Reduced HELOAN and HELOC rate update time by 10x through an internal management platform (ASP.NET Core + MySQL), eliminating manual record adjustments and spreadsheet-based tracking.
- Facilitated development of a hackathon-winning platform (Angular + Salesforce) to streamline appointment scheduling with members and Certified Financial Specialists directly.
Solutions Architect Intern
Automated big data management and supported sales meetings by gathering client requirements.
- Worked with solutions architects, enterprise clients, and storage engineers to modernize ONTAP-based backup workflows for Oracle databases.
- Automated Oracle RMAN backup processes using the Oracle SDK and Python, decreasing backup execution time by 50% and minimizing manual DBA intervention.
- Designed a Python and Bash-based monitoring and alerting system to track storage array health metrics, cutting manual monitoring time by 90%.
User Interface Engineer
Developed testing oracles to ensure the front end of the web application performs as expected. Engineered pipelines to ensure issues were prioritized and resolved efficiently.
Projects
Featured
More Projects
Task Scheduling
Research on CPU and GPU scheduling algorithms, with a focus on AI workloads and hybrid CPU-GPU scheduling strategies.
A project that breaks Dijkstra's algorithm, a shortest path algorithm for graphs. Based on the paper 'Breaking the Sorting Barrier for Directed Single-Source Shortest Paths'.
Ares
SOC2 compliance platform leveraging AI and deep cybersecurity technology launched on VSCode Marketplace.
Molecule Mutation Prediction
Support vector machine to identify compounds that inhibit BRAF V600E mutation using PubChem chemical data.