Divyang Chauhan
SUMMARY
Applied AI Engineer with 5+ years of building event-driven services and multi-tenant platforms. Developed provider neutral, tool-using agent systems with context isolation and evals. Led a 9-person team shipping security software for multinational banks. OSCP certified.
PROJECTS
Pramana: Multi-agent smart-contract vulnerability scanner
- Designed provider-neutral three agent system with context isolation and tool usage that produces executable PoC
- Built an eval harness covering 14 labeled vulnerabilities in 11 classes plus a patched negative control, with baselines, cost/latency tracking, CI and offline tests
- Benchmarked the agents across Claude, GPT, and Kimi; best runs found 13/14 vulnerabilities with 0 confirmed false positive
ClinchCV: Full-stack AI resume analysis and job targeting
- Shipped a full-stack AI product that parses resume PDFs and gives structured output, rubric-based scoring, ATS checks, job-fit analysis, and contextual rewrites
- Enforced structured outputs with schema validation, automatic retries for invalid output, and model fallback
PROFESSIONAL EXPERIENCE
Kleros, RemoteFeb 2024 – May 2026
Backend Engineer
- Decomposed a monolithic NestJS backend into independently scalable API and automation services; established the monorepo and migrated 57K+ blockchain events across databases with ~1 minute of downtime
- Architected a NestJS EVM event-ingestion service processing events across Ethereum, Gnosis, and Arbitrum for 23+ downstream consumers, including notification workflows serving 700+ active jurors
- Automated dispute progression and juror staking across networks; used transaction simulation and batched calls to improve execution reliability and reduce gas usage by ~22%
NST Cyber, RemoteJan 2023 – Feb 2024
Software Engineer Team Lead
- Led a cross-functional team of 9 building Assure v2, a multi-tenant vulnerability-triage and threat-surface management platform sold directly to multinational banks and white-labeled by cybersecurity resellers
- Built an AI-assisted APT attribution workflow using the OpenAI API to map threat-surface to threat actors
- Replatformed Assure from OutSystems, eliminating ~$200K/year in platform costs while enabling 3-level tenancy, data isolation modes, and white-labeling; authored ~70% of the backend and shipped in six months
- Wrote Terraform for Assure v2's core AWS infrastructure, reducing deployment time from ~2 hours to 10–15 minutes; introduced Cypress E2E tests, reducing full regression from hours to under 30 minutes
Software EngineerMar 2021 – Jan 2023
- Delivered Assure v1 concept to production in 3 months, authoring 80% of codebase and onboarding first 5 enterprise clients; rapid adoption exceeded platform limits, driving decision to rebuild on scalable architecture
- Developed Tigress's Django REST Framework backend from scratch; designed a MySQL schema supporting 10M+ records and created a serializer abstraction that reduced new endpoint development time by 60%
- Eliminated distributed-task bottleneck in Tigress's scanning engine by architecting cross-instance Celery dispatch, reducing scan time by 75% and scaling the platform from 10 to 100 daily scans without infrastructure changes
CERTIFICATION & EDUCATION
Offensive Security Certified Professional, OffSec2017
IIT Bombay, B.Tech coursework, Metallurgical Engineering & Materials Science2016
SKILLS
AI Systems: OpenAI, Anthropic, agent orchestration, tool calling, structured outputs, context isolation, evals
Languages & Databases: Python, TypeScript, JavaScript, Solidity, SQL, PostgreSQL, MySQL, MongoDB
Frameworks: Node.js, NestJS, Django REST Framework, Pydantic, GraphQL, React, Next.js, Foundry, Slither
Infrastructure & Messaging: AWS, Docker, Terraform, Celery, Kafka, RabbitMQ