Prajwal Umesh
Nagalakshmi
Full-Stack Software Engineer | AI Developer
About Me
I build scalable systems and AI agents. MSCS from CSU Dominguez Hills (GPA 3.7). I've architected distributed systems load-tested to 200K+ concurrent requests (DealCart+), built AI security SDKs with LangGraph and threat detection (Agent Sentinel), and engineered high-performance KV stores achieving 104K+ QPS in C++ (NanoKV). Previously at Appleute, I reduced checkout times by 60% and boosted engagement by 40%. I also volunteer as a Full Stack Developer at Saayam For All, contributing to GenAI, including migrating legacy LLM code to LangChain across microservices and refactoring monolithic Lambdas into class-based API layers.
Experience
- Migrated legacy LLM code to LangChain across 3+ Python microservices, improving modularity and maintainability of GenAI features
- Refactored monolithic Lambda functions into class-based API layers across endpoints, enforcing separation of concerns
- Engineered Betra, an AI-powered cashierless shopping app that reduced retail checkout times by 60% for end-users
- Propelled user engagement by 40% by optimizing React Native rendering speeds via lazy loading and component memoization
- Slashed API response latency by 25% by implementing NestJS backend caching strategies and optimizing database queries
- Integrated internationalization (i18n) and secure authentication flows, enabling multi-language support and offline access
- Mentored 30+ students in quant fundamentals, improving pass rates by 15% through adaptive teaching strategies
Leadership and Awards
- Spearheaded the development of a multi-agent orchestration system integrating Finance, Travel, and Media retrieval agents
- Adopted the MCP standard to enable modular, context-aware coordination and seamless tool hand-offs between agents
- Facilitated weekly Python programming sessions for 15+ students, delivering instruction based on Stanford's CS106A curriculum
- Coached a global cohort in debugging and algorithmic thinking, resolving technical roadblocks to ensure concept mastery
Projects
Featured Projects
- Architected Java 21 microservices using gRPC, Protobuf, and SSE to enable low-latency, real-time streaming
- Orchestrated system reliability via parallel fan-out, rate limiting, SAGA rollbacks, and distributed tracing
- Devised a custom auto-scaling solution (1→5 instances) triggered by real-time CPU/memory metrics and error rates
- Validated system stability under 200K+ concurrent requests (P95 <250ms) via rigorous load testing on AWS EC2
- Built an AI Agents security SDK supporting 100 concurrent sessions and detecting 20+ threat types (XSS, SQL/Prompt injection)
- Implemented a 5-agent LangGraph workflow to automate security analysis and reporting via OpenAI integration
- Launched a full-stack observability platform (FastAPI/Next.js) with real-time streaming and W&B distributed tracing
- Designed a thread-safe global event registry for multi-agent state aggregation in distributed systems
- Engineered a multithreaded key-value engine achieving 104,000+ QPS and <1ms latency, validated via custom load tools
- Implemented a cross-platform Event Loop using epoll (Linux) and kqueue (macOS) to scale to 10k+ concurrent connections
- Designed a custom binary protocol to replace JSON, reducing parsing overhead by 90% through zero-copy buffering
More Projects
Model distillation: compiles expensive gpt-4o reasoning into cheap fine-tuned specialists
AI-powered code scanner that identifies technical debt and forecasts future impact
AI release notes from PR diffs for devs and marketers
Networking app with business card/LinkedIn QR scan + AI follow-ups
Agentic resume analyzer with iterative improvement and anti-fabrication
Technical Skills
Let's Create Something Amazing
Ready to Connect
I'm actively seeking new graduate software engineering opportunities. Let's discuss how I can contribute to your team's success and build the future together.