Himanshu Kiran Garud

About

Backend/ML engineer who has built both enterprise data pipelines and real-time AI products: modular Python (Flask/FastAPI) services integrating automation scripts and low-latency ML inference (MediaPipe, PyTorch) plus OpenAI-powered feedback. Demonstrated measurable performance wins (~30% faster HR workflows; ~40% faster AWS pipelines across 100+ Oscar Health feeds) and strong multi-tenant/data-isolation patterns (schema-based isolation, RBAC, microservices).

Experience

Education

Key Strengths

• Designed modular Python/Flask backend with stable endpoints and fallback/cached responses for unreliable automation scripts
• Improved internal HR workflow backend performance by ~30% via WAL mode, batched writes, and retry logic
• SQLAlchemy/PostgreSQL performance tuning using access-pattern-driven schema design, composite indexes, and materialized views
• Built real-time ML backends (FastAPI) with low-latency inference using multithreading, batching, and model warmup
• Integrated GPT/OpenAI API layer to generate personalized feedback from ML outputs
• Implemented schema-based data isolation and RBAC/scoped queries for multi-tenant-style systems
• Optimized high-throughput background pipelines on AWS (100+ production feeds) with parallel modular tasks, caching, and retries; ~40% faster processing
• Designed and evolved backend data processing system supporting 100+ production data feeds for US healthcare clients
• Restructured monolithic workflows into modular microservices (Spring Boot/Python) on AWS and GCP
• Led production migration from GraphQL services to optimized SQL-driven microservices with zero downtime via parallel run and feature-flag rollout
• Improved API response time by ~30% through query and architecture changes
• Increased distributed pipeline throughput by ~40% via better orchestration and monitoring
• Strong data integrity and risk controls (checksum validation, automated reconciliation, schema validation)
• Built reliable low-latency FastAPI services using Pydantic schemas, async endpoints, and predictable error handling
• Implemented layered, evolvable security (JWT auth, RBAC, row-level scoping, centralized middleware, rate limiting, logging)
• Identified subtle ETL failure mode (partial datasets passing schema checks) and added completeness/time-window validations to prevent downstream corruption

Reference Highlights

• Strong backend technical depth
• Designs clean, modular backend components
• Writes reliable production-grade Python code
• Clear reasoning about API structure and data flow
• Good Flask practices beyond basics
• Strong SQLAlchemy/ORM skills
• Optimizes database queries effectively
• Balances speed and maintainability without overengineering
• Designs scalable, maintainable architectures under startup pressure
• Breaks complex workflows into clean, testable services
• Comfortable making decisions and iterating quickly
• Adapts designs as requirements evolve
• Very effective cross-functional collaborator
• Clear communicator who keeps teams aligned
• Proactive about sharing progress and clarifying requirements
• Approachable and solution-oriented
• Calm and methodical under tight deadlines
• Finds root causes (not just symptoms)
• Delivers clean fixes without introducing new issues
• Verifies fixes end-to-end
• Reliable teammate
• Strong technical depth in Python
• Writes clean, well-structured, maintainable code
• Thoughtful software development with strong fundamentals
• Designs modular, scalable, easy-to-extend components
• Balances fast iteration with clean engineering practices
• Breaks down complex workflows into manageable, reusable pieces
• Collaborates extremely well in fast-moving AI environments
• Clear communicator across faculty, peers, and students
• Adapts quickly as project directions shift
• Calm and focused under pressure
• Efficient, thorough debugging and problem-solving
• Steps up to coordinate and deliver working end-to-end systems for deadlines