Krishna M.

Key Skills

Software

CVAT

Snorkel AI

Prodigy

Label Studio

Top Subject Matter

LLM evaluation

RLHF-style alignment verification

red teaming for reasoning agents

Top Data Types

Text

Document

Computer Code Programming

Top Task Types

Red Teaming

Question Answering

RLHF

Function Calling

Computer Programming/Coding

Evaluation/Rating

Freelancer Overview

AI Verification Contractor (Fractional). Brings 5+ years of professional experience across complex professional workflows, research, and quality-focused execution. Education includes Professional Diploma in IT, N/A and Bachelor of Business Administration, N/A. AI-training focus includes data types such as Text and labeling workflows including Evaluation, Rating, and Red Teaming.

Labeling Experience

Founder & Solutions Architect — Training-style curation and validation protocols (2025–Present)

TextPrompt + Response Writing (SFT)

Authored precise prompt-response pairs to enable instruction tuning and high-quality supervised learning-style training. Developed detailed justification rubrics explaining why outputs failed semantic or structural criteria. Used these curated training artifacts to improve multi-step reasoning alignment. • Wrote high-density prompt-response pairs for complex tasks • Defined rubric criteria for semantic and structural failure • Ensured strict reasoning/constraint alignment in curated outputs • Accelerated model alignment using documented failure reasons

2025 - Present

AI Training Specialist & Verification Contractor (Fractional) — Adversarial Prompting & Red Teaming (2025–Present)

TextRed Teaming

Conducted systemic red teaming by engineering adversarial prompts and analyzing model behavior under stress. Identified vulnerabilities, hallucination patterns, and edge-case errors in reasoning-focused agents. Supported reinforcement learning from human feedback (RLHF) and safety alignment efforts through structured findings. • Generated adversarial prompts for stress tests • Detected hallucination and edge-case failure modes • Mapped vulnerabilities and safety alignment leaks • Reported findings to guide alignment improvements

2025 - Present

AI Training Specialist & Verification Contractor (Fractional) — Independent Technical Advisory (2025–Present)

Text

Performed audits of multi-turn LLM reasoning loops, checking responses for logical consistency, constraint adherence, and factual precision. Produced structured evaluation rationales indicating why specific outputs failed semantic or structural benchmarks. Used these findings to support model alignment and training-quality improvements. • Evaluated logical consistency across multi-step conversations • Verified truthfulness and factual precision in dense outputs • Assessed adherence to constraints and procedural correctness • Documented failure reasons using detailed rubrics

2025 - Present

Education

Work History