.png){kind=logo}
KS
Keerthana Senthilnathan
Junior Machine Learning Engineer specializing in LLM systems and inference reliability
California, USAOpen-Source Contributor — ML Health & Inference Reliability1 years experienceJuniorTechnologyArtificial IntelligenceCloud Computing
ScreenedIdentity Verified
Connect with Keerthana
Keerthana already has a relationship with Reval, so a warm intro from us gets a much better response than cold outreach.
Recommended
Already have an account?
About
ML/LLM infrastructure-focused engineer who built a production stateful LLM inference service that cuts latency and GPU compute for repeated/overlapping prompts via caching with correctness guardrails. Strong in Kubernetes-based deployment and reliability engineering, using A/B testing and similarity-based evaluation to quantify performance gains without sacrificing output quality.
Experience
Open-Source Contributor — ML Health & Inference Reliabilityllm-d
Software Development Engineer Intern — Cross-Border TransportationAmazon.com Services LLC
Research InternCarnegie Mellon University, Xu Lab
Generative AI InternSamsung R&D Institute
Education
University of California, San Diegomaster, Data Science (Artificial Intelligence & Machine Learning) (2026)
National Institute of Technology, Tiruchirappallibachelor, Electronics & Communication Engineering (2024)
Key Strengths
- Built and deployed a stateful LLM inference service to reduce latency and GPU cost for overlapping prompts
- Balances correctness vs performance using similarity thresholds and automatic cache bypass
- Strong production reliability focus (bounded caches, eviction strategies, memory pressure monitoring to prevent OOM)
- Metrics-driven evaluation using controlled A/B experiments (latency, cache hit rate, GPU savings, output similarity distributions)
- Hands-on container orchestration in Kubernetes (resource limits/requests, scheduling, restart policies, horizontal scaling)
- Designs AI agents as observable, composable, idempotent stages with measurable contracts and guardrails
- Pragmatic model/retrieval/prompting selection driven by constraints and empirical validation (offline eval + A/B tests)
Discover more candidates like Keerthana
Search across thousands of pre-screened, high-quality, high-intent candidates on Reval.
Search Talent