Soham Bhattacharya

Bachelor of Arts in Computer Science and Economics
Minor in Policy and Business Management

• GPA: 3.6/4.0 · ACT: 36/36
• Coursework: Operating Systems, Political Economy, Artificial Intelligence, Macroeconomic Theory, Computer Security.
• Activities: Tech Correspondent for NYU Econ Review; Infrastructure and Research for NYU OSIRIS (Cybersecurity Lab).

• Engineered a pipeline to generate rightsized, least-privilege permission sets across 32k+ first- and third-party apps and AI agents.
• Scaled the Azure Detection Platform pipeline to all 1.5k+ Microsoft Graph workloads by mapping endpoints with agentic code analysis.
• Reduced risk ratings of compromised Entra identities by 30%, limiting the attack vector from exploits by agentic actors.

• Built a custom AI-driven firm management platform automating email, outreach, and due diligence workflows.
• Reduced firm operations from multiple weeks to hours via an automation pipeline using FastAPI, React, RAG, and SERP APIs.
• Authored technical assessments for prospective investments, benchmarking differentiation and scalability for over $5M in deal value.

• Released the first GDPR-compliant search platform across 800M+ candidate profiles, improving match rate 30x (1% to 30%).
• Built and deployed a containerized FastAPI/Pinecone API with custom RAG context, an LLM harness, and a Next.js frontend.
• Scaled the platform to 5,000+ users and several business clients; secured pre-seed investment from LvlUp Ventures.

• Built onboarding automation for all 1.5k+ devices across 12 global data centers using PostgreSQL, FastAPI, and Prometheus.
• Engineered a vulnerability tracking API and platform, identifying and surfacing 28K+ threats across all control center hardware.

• Migrated 3K+ OS build packages from NAS to Artifactory across all Cisco regions, enabling caching for global build infrastructure.
• Rewrote the SDK delivery pipeline, cutting 5 redundant processes, storage demand by 1TB+ per release, and build time by 10%.

• Built a ZK authentication demo using Rust, Tokio, and ZK-SNARKs that was used to close a $2M round; presented the architecture at ETHDenver.

Engineered an adversarial multi-agent evaluation platform in a sandboxed VM to test lateral movement and permission escalation under competitive pressure; instrumented traces to surface emergent behaviors including cooperation, replication, and communication.

Recompiled Pokémon DS games for Windows, macOS, and iOS. First in the world to deliver native releases for 6 out of 7 games. Also ported to the web with optimized WASM, new multiplayer implementations, and server-side saves.

Benchmarked fruit fly and zebrafish connectomes against RNNs across control and temporal tasks, varying linear/MLP encoders and decoders to test whether connectome structure confers task-specific advantages.

“Methods and Systems for Selecting an Optimal Proof System for Zero-Knowledge and Other Proofs”

“SyncTCI: Synchronized Context Injection on Distributed Real-Time Inference Hardware”

“Gendered Impact of High-School Financial Literacy Requirements on Young Adults’ Credit Behavior”

“Powering the AI Revolution: Inside the Energy Economics of Data Center Development”

“Inside the Costs of Compliance, and the Downstream Effects of Delve’s Fraud”

• Hack-Nation — Winner, VC Track: Over 6,000 participants; won with Doppio Labs and received admission to a venture incubation program.
• A16Z Headstarter GLW Hackathon — Finalist: Top 5 out of over 3,000; pitched Doppio Labs as a lead-generation tool.
• Los Altos Hacks V — 2nd Place Overall: Developed a natural language and OCR system to generate notes from text, before ChatGPT.
• NYU Dean’s List: College of Arts and Sciences award for academic achievement.

Languages: Python, Dart, Go, Rust, C/C++, KQL/SQL

Frameworks & tools: FastAPI, PostgreSQL, Apache Spark, Kubernetes, Docker, Git, Pinecone, Azure Data Lake, OpenAI Evals, Entra ID

Topics: Agent Identity, Neural Networks, Reinforcement Learning, Post-Training, AI Security, Agent-Driven Threats, AI Alignment

Interests: Music Composition & Production, Photography, Film Development, Labor Economics, Psychology