Career Advisor Backend
RAG-powered resume feedback engine using FastAPI and Qdrant. Built deterministic pipelines for chunking, embeddings, ranking, and caching to deliver reliable LLM-driven career guidance.
I build high-scale distributed systems and AI/LLM-powered backend services that handle 10× transaction growth, deliver 8× throughput, and drive $16M+ in revenue.
I'm Sandesh — a Backend Engineer at First Help Financial, building production backend services and AI/LLM-powered systems for a high-volume auto-finance loan origination platform. I hold a Master's in Computer Science from Northeastern University (Khoury College) and have 4+ years of experience across First Help Financial, Thoughtworks, and TCS, building distributed .NET Core microservices on Azure, serverless workflows on AWS, and enterprise banking APIs with Spring Boot.
I specialize in microservices, cloud-native architectures, RAG/LLM systems, reliability engineering, and observability. I care deeply about clean code, test-driven development, and systems that don't break at 3 AM.
Own end-to-end design and delivery of production backend services, authoring architecture decision records (ADRs) to align technical direction across the team. Develop and extend core services for the loan origination system powering high-volume auto-finance workflows. Drive an AI-first development workflow, orchestrating Cursor with structured plan/execute prompting and memory-file context management, cutting delivery time per ticket by 35% while improving code consistency. Lead a RAG-based agentic AI proof-of-concept to automate underwriting document review in the loan origination flow, applying retrieval-augmented pipelines and tool-calling.
Implemented serverless backend services using AWS Lambda, API Gateway, Step Functions, and DocumentDB for high-volume financial workflows. Built a reproducible AWS-local dev environment (SAM + LocalStack) reducing debugging time by 55%. Developed Python libraries for structured logging and automated fault-handling, achieving 95%+ test coverage.
Built distributed .NET Core microservices on Azure. Scaled an ordering platform to 8× throughput and 10× global transactions. Built a dev onboarding portal cutting integration time by 60% and enabling $16M+ revenue. Boosted performance 3× via profiling, parallel execution, and optimized SQL.
Implemented backend for enterprise banking systems using Spring Boot, JPA/Hibernate, and SQL. Improved API performance by 20% via query optimization. Developed Angular UI components integrated with Spring Boot APIs.
RAG-powered resume feedback engine using FastAPI and Qdrant. Built deterministic pipelines for chunking, embeddings, ranking, and caching to deliver reliable LLM-driven career guidance.
Full-stack job matching platform with RBAC, secure auth, and optimized PostgreSQL queries. Efficient pagination, index-driven search, and prepared statements for fast responsiveness.
Modular engine supporting sharpening, blurring, and color correction. Factory, Builder, Visitor, Adapter, and Command patterns with concurrent transformation pipelines.
.NET Core REST API with Entity Framework Core, Swagger docs, Fluent Validation, AutoMapper, and CI/CD via GitHub Actions. Comprehensive unit test coverage.
Custom maze environment where multiple RL agents (Q-Learning, etc.) navigate from start to exit, optimizing for shortest path. Modular agent architecture with visualization.
"I have worked with Sandesh for around 2 years, and he has been an excellent engineer throughout. From the get-go, he started owning features and delivering them. His analytical skills are very good. He would be an asset in any team he works."
"Sandesh is technically strong and keen to learn new technologies. He always brings his points in any discussion. He has very good communication skills and took sessions for the team on multiple topics."
"Sandesh possesses the qualities and skills necessary not only to excel in a rigorous academic environment but also to make meaningful contributions to any community he joins."
Whether you have a project, want to talk backend systems and AI/LLM engineering, or just want to connect — I'd love to hear from you.