Hemanth G.

About Hemanth G.

I’m a results‑driven Software Engineer with 4+ years of experience building scalable cloud applications, ML pipelines, and data‑driven systems across fintech, enterprise, and geospatial domains. I specialize in designing end‑to‑end solutions that turn complex data and modern AI capabilities into real business impact.My background spans AWS, GCP, Python, Java, Spring Boot, PyTorch, TensorFlow, vector databases, and distributed systems, with a strong focus on GenAI, MLOps, and cloud‑native architectures.Key Highlights:• Architected a GenAI news‑curation platform at Fannie Mae that reduced Credit Analyst research time from ~3 hours to near‑zero by automating daily risk‑analyzed insights using a scalable RAG pipeline on AWS.• Improved vector retrieval performance by 20–30% by integrating PgVector into Aurora Postgres and optimizing embedding‑based search.• Enhanced ML model accuracy by 83% through U‑Net–based satellite image segmentation using PyTorch and TensorFlow.• Delivered insights from data records using Scikit‑learn and advanced data processing techniques.• Engineered cloud‑native applications serving 200+ daily users with Spring Boot, JavaScript, and GCP.• Automated CI/CD pipelines with Jenkins, Git, Docker, and Kubernetes, reducing deployment cycles by 20%.With a Master’s in Computer Science from the University at Buffalo, I bring a strong foundation in machine learning, cloud architecture, and backend engineering — along with a passion for building systems that are scalable, reliable, and genuinely useful.If you’re looking to connect with someone who can design modern cloud/ML solutions, optimize complex pipelines, and deliver measurable outcomes, I’d love to chat.

Experience

1. Software Developer

   Fannie Mae

   Mar 2025 — Present · Reston, VA, US

   Architected an end‑to‑end GenAI news‑curation platform that reduced Credit Analyst research time from ~3 hours to near‑zero, by designing a scalable RAG pipeline on AWS leveraging Step Functions, Lambda, SQS/SNS, and Aurora Postgres.• Engineered high‑performance vector retrieval, improving accuracy and reducing latency by 20–30%, by integrating PgVector into Aurora Postgres and optimizing embedding‑based similarity search.• Automated news discovery for 700+ business partners, eliminating manual keyword searches and accelerating risk‑insight generation, by orchestrating ingestion, ranking, and summarization workflows powered by Claude Sonnet 3.5/3.7.• Orchestrated four independent downstream delivery channels, strengthening system reliability and throughput, by implementing decoupled event‑driven architectures using SQS/SNS and modular orchestration patterns.• Enhanced operational resilience and reduced manual intervention by 40%, by implementing CloudWatch dashboards, alerting strategies, and AWS Batch–based scheduled workflows.• Optimized ingestion‑to‑delivery performance, boosting pipeline throughput, by refining Lambda execution paths and standardizing parallel Step Function state machines for high‑volume processing.

Education

• University at Buffalo

  Master of Science - MS, Computer Science

• Sri Venkateswara University

  Bachelor of Technology, Computer Science and Engineering

  2016 — 2020