Samarth Garg

ML Engineer Intern at Manthhan (2026-01 – Present)

• Resolved GitHub Actions CI failures during Python 3.12 to 3.14 migration by optimizing PR check workflows and configuring 3 GB swap memory to overcome AWS CodeBuild storage constraints, ensuring stable package installation and test execution.
• Decreased Docker image size from 3.23 GB to 1.73 GB (47% reduction), reducing AWS Elastic Container Service overhead.
• Improved document classification accuracy from 32% to 95% by optimizing regex-based pattern extraction on OCR-processed text, increasing the robustness of rule-based classification.
• Engineered an LLM-driven document processing system that reduced manual reading effort by 70%.

AI Engineer Intern at YourAllyStack (2025-09 – 2025-12)

• Built an end-to-end interview evaluation pipeline that converted S3 video interviews into transcripts and structured Q&A pairs (+/-1 error margin) with automated student and rubric-based admin evaluation.
• Optimized Q&A pair extraction from 1-hour transcripts to under 1 minute by implementing strategic text chunking, enabling efficient processing for both human-conducted and AI-conducted interview evaluations.
• Enabled scalable rubric-driven LLM evaluation by parsing Excel rubrics into dynamic scoring indicators across interview sessions, reducing token cost by 60% and latency by 80%.
• Integrated database-backed storage for transcripts, scores, and feedback for efficient retrieval and auditing.

B.Tech (Hons.) in Computer Science and Engineering with Specialization in Artificial Intelligence and Data Science – LNM Institute of Information Technology, Jaipur, Rajasthan (2022-10 – 2026-05)