Senior Software Engineer · AI Security Researcher · Georgia Tech M.S. CS (AI)
I build high-performance distributed systems and research what breaks when autonomous agents start making security decisions together. Spent years hardening consensus on a live sharded blockchain in Rust — now applying those instincts to multi-agent LLM threat models.
| 🔐 Research | Consensus security in multi-agent LLM systems — BFT quorum design, adversarial hardening, prompt injection resistance for autonomous agents in high-stakes CyberSec environments |
| 🦀 Engineering | Distributed Systems Engineer @ Liberdus — Rust, TypeScript, Node.js on a live sharded blockchain |
| 🎓 Academia | M.S. Computer Science (AI) @ Georgia Tech, School of Cybersecurity and Privacy |
Languages Rust · TypeScript · Python · Go · Solidity · Kotlin
Runtime Node.js · gRPC · Protocol Buffers · WebSocket
Infra Docker · Kubernetes · PostgreSQL · RocksDB
Blockchain EVM · Web3.js · Shardus · Farcaster
ML / Data PyTorch · scikit-learn · XGBoost · Pandas
Security Static Analysis · PCAP · Wireshark · Bug Bounty
Consensus Security in Multi-Agent LLM Systems
Securing multi-agentic AI systems in Cybersecurity Operations Centre (SOC) workflows to defend against adversarial attacks. Covers BFT quorum design, insider-agent threat models, prompt injection resistance, and verifier-based trust models. Spans SIEM, SOAR, Case Management, Agent EDR, Auditing and Shadowing.
AI-based Market Intelligence for Farmer Collectives
Collaborative research with IIT Delhi, BITS Pilani, CCD, and Gram Vaani. Production Android app with integrated ML forecasting deployed for rural agricultural markets. Field-validated with real market data. Presented at ACM COMPASS '23.
Published · ACM JCSS · Funded by Google AI for Social Good
Security on Low-Cost Android Devices
Static and dynamic analysis, PCAP-based network traffic analysis via Wireshark, malware and bloatware identification across device tiers.
HCI & Security Lab · BITS Pilani
| Project | Proof | Contribution | Stack |
|---|---|---|---|
| Liberdus / TSS Signer | PR #5 | Multi-network event processing, gas estimation, adaptive batch sizing, rate limiting, block-state persistence, Ed25519 request validation | Rust · TypeScript · Node.js |
| Farcaster / Snapchain | Issue #544 | Timestamp and block-based event filtering APIs; serialization optimizations for network and database performance | Rust · RocksDB · gRPC |
| Cosmos EVM | External contribution | Height-keyed caching for blocks, block results, and fee params to eliminate redundant RPC calls | Go · JSON-RPC |
| Shardeum Mainnet | Open-source production work | Consensus hardening, node monitoring, networking latency improvements, cryptographic performance optimization, transaction authorization, service tests | Rust · TypeScript · Node.js |
| Shardus | Open-source production work | Distributed ledger components for sharded consensus, failed-node detection, account storage, executable caching, reverse proxies, validator-node communication | TypeScript · Node.js |
| Project | What it proves | Stack |
|---|---|---|
| SOC Defender | AI-assisted SOC alert triage, evidence collection, retrieval-grounded security workflows | Python · LangGraph · RAG · Wazuh · Qdrant |
| Shardeum MCP Server | MCP tooling that exposes Shardeum blockchain operations to AI agents and developer workflows | TypeScript · MCP · Blockchain |
| Multi-Agentic BFT | Byzantine fault-tolerant quorum voting for reliable multi-agent decision making in untrusted environments | Python · Multi-Agent Systems · BFT |
From production work on a live sharded EVM blockchain.
- 34% networking latency reduction — 82ms → 54ms, further cut 29% with Rust libsodium
- 14× ecrecover address verification speedup — 10ms → 700µs
- 60% RPC throughput increase via WebSocket-based data distribution
- 10,000+ req/sec sustained on node monitoring system
- Critical vulnerabilities found and reported in Shardeum & Shardus bug bounty programs
- ACM Journal on Computing and Sustainable Societies (JCSS) — AI-based Market Intelligence Systems for Farmer Collectives
- Funded by Google AI for Social Good
- Invited to present at ACM COMPASS '23
- ACM track publications in progress (Georgia Tech affiliation)




