Amit Kushwaha

Key Skills

Software

Snorkel AI

Data Annotation Tech

Mindrift

Toloka

Top Subject Matter

LLM evaluation and coding task benchmarking

LLM evaluation and benchmarking

Top Data Types

Text

Top Task Types

Segmentation

Text Summarization

Text Generation

Fine Tuning

Evaluation Rating

Function Calling

Computer Programming Coding

Freelancer Overview

LLM Trainer (Freelance). Brings 1+ years of professional experience across complex professional workflows, research, and quality-focused execution. Core strengths include Snorkel AI, Internal, and Proprietary Tooling. Education includes Bachelor of Technology, Indian Institute of Information Technology Ranchi (2022). AI-training focus includes data types such as Computer Code and Programming and labeling workflows including Evaluation and Rating.

Labeling Experience

LLM Trainer (Freelance)

Snorkel AI

As an LLM Trainer at Snorkel AI, I reviewed and validated coding tasks generated from open-source pull requests. I ensured structure, solution correctness, test case sufficiency, and metadata accuracy for high-quality benchmarks used in AI agent evaluation. I evaluated coding agents using Docker-based oracle and NOP test pipelines and submitted comprehensive review reports to enhance dataset reliability. • Assessed task validity, instruction clarity, and alignment with test requirements. • Analyzed logs for agent outcomes across models like GPT-5 and Claude Sonnet. • Conducted failure mode and timeout troubleshooting to maintain benchmark standards. • Leveraged Docker and internal/proprietary validation workflows for end-to-end pipeline execution.

As an LLM Trainer at Snorkel AI, I reviewed and validated coding tasks generated from open-source pull requests. I ensured structure, solution correctness, test case sufficiency, and metadata accuracy for high-quality benchmarks used in AI agent evaluation. I evaluated coding agents using Docker-based oracle and NOP test pipelines and submitted comprehensive review reports to enhance dataset reliability. • Assessed task validity, instruction clarity, and alignment with test requirements. • Analyzed logs for agent outcomes across models like GPT-5 and Claude Sonnet. • Conducted failure mode and timeout troubleshooting to maintain benchmark standards. • Leveraged Docker and internal/proprietary validation workflows for end-to-end pipeline execution.

2026 - 2026

LLM Trainer (Freelance)

As an LLM Trainer at Dusker AI, I developed and maintained Terminal-Bench tasks to systematically evaluate LLMs. I designed task specifications, executed Dockerized test pipelines, and validated large-scale benchmarks using Oracle and NOP agents. I assessed model correctness and ensured robust and reproducible evaluation environments. • Evaluated agent performance across classification, automation, and ML-driven domains. • Designed and executed reproducible validation pipelines for consistent results. • Oversaw Oracle agent resolution and ensured baseline agent failures as expected. • Used internal/proprietary tools and Docker for validation and benchmarking processes.

As an LLM Trainer at Dusker AI, I developed and maintained Terminal-Bench tasks to systematically evaluate LLMs. I designed task specifications, executed Dockerized test pipelines, and validated large-scale benchmarks using Oracle and NOP agents. I assessed model correctness and ensured robust and reproducible evaluation environments. • Evaluated agent performance across classification, automation, and ML-driven domains. • Designed and executed reproducible validation pipelines for consistent results. • Oversaw Oracle agent resolution and ensured baseline agent failures as expected. • Used internal/proprietary tools and Docker for validation and benchmarking processes.

2025 - 2025

Education

Work History