Building high-throughput, deterministic software where performance, algorithmic correctness, and scale are non-negotiable. Experienced in orchestrating governed AI agents, authoring constraint satisfaction solvers with sub-second convergence, and enforcing zero-leakage financial invariants.
Explore production systems spanning Agentic AI pipelines, constraint satisfaction solvers, fault-tolerant financial saga engines, and high-throughput evaluation frameworks.
Agentic AI platform for proactive customer intelligence and autonomous task execution. Detects user intent, generates multi-step execution plans, and executes verified tool adapters across CRM, Jira, ServiceNow, and financial ledgers with safety guardrails and immutable audit logs.
Engineered with rigorous separation of concerns. Includes E2E sandbox evaluation suites and interactive demonstration interfaces for enterprise walkthroughs.
Production-grade constraint satisfaction engine scheduling 800 students × 35 companies across 20 rooms over 4 days. Guarantees zero double-bookings with sub-second convergence. Features real-time disruption replanning with local-repair heuristics.
Resilient against 4 disruption classes: student delays, interviewer dropouts, company withdrawals, and room failures. Version-controlled state management with zero-loss rollback.
Autonomous revenue recovery platform with governed AI agents. Analyzes payment failures across a 6-category taxonomy, generates AI recovery strategies, enforces OPA Rego policy firewall checks, and orchestrates durable workflows via Temporal with cryptographic ledger settlement (immudb).
Built for Razorpay AI Buildathon. Integrates AI Strategy Advisor, Open Policy Agent (OPA) Firewall, Temporal Saga Orchestrator, and immudb cryptographic ledger.
LLM metacognition evaluation framework measuring AI confidence calibration, error detection, belief correction, and epistemic state recognition. Exposes failure modes missed by MMLU & TruthfulQA.
High-frequency payment settlement optimizer with priority queues and deadline-aware scheduling. Predicts gateway congestion using ML models with adaptive retry backoff.
Machine learning candidate scoring system utilizing XGBoost and NLP feature vectors for automated profile evaluation.
Sub-second TTI landing platform optimized for high conversion, zero CLS, and 100/100 Lighthouse performance metrics.
Vehicle rental management system supporting real-time fleet tracking, transactional booking consistency, and billing.
Multilingual agricultural decision engine integrating OpenWeatherMap, ISRO Bhuvan satellite feeds, and E-NAM market data.
Client-side AES-256 encrypted note application paired with real-time threat monitoring and WebSocket alert streams.
LangChain-driven RAG pipeline backed by Pinecone vector storage with Whisper local speech-to-text and proactive PII masking.
Personal healthcare management portal with automated hospital indexing, pharmacy directory, and expense estimators.
Core technical skills organized across languages, agentic workflows, distributed backends, cloud infrastructure, and high-performance frontend interfaces.
Practical engineering experience delivering production-grade web systems and building distributed tools.
I am targeting SDE-1 / Software Engineering roles at Google and leading technology companies. If you are looking for an engineer who thrives on algorithmic challenges, deterministic execution, and agentic workflows, let's connect.