Shivam K.

Key Skills

Software

OneForma

Top Subject Matter

Large Language Model (LLM) alignment and evaluation

Search relevance annotation and dataset quality control

Multimodal training dataset labeling (text/audio/image)

Top Data Types

Text

Audio

Image

Top Task Types

RLHF

Classification

Freelancer Overview

AI Trainer & Language Model Evaluator (Contractor). Core strengths include Outlier, OneForma, and CrowdGen. Education includes Bachelor of Technology, N/A (2024). AI-training focus includes data types such as Text and Audio and labeling workflows including RLHF, Evaluation, and Rating.

Labeling Experience

AI Trainer & Language Model Evaluator (Contractor)

TextRLHF

Served as an AI trainer and language model evaluator, using RLHF workflows to score and rank LLM response sequences. Built supervised fine-tuning (SFT) ground-truth interaction datasets by authoring error-free text pairings across technical and conversational verticals. Performed adversarial prompt simulation to identify, document, and patch alignment blind spots, bias, and factual hallucinations. • Evaluated factual precision, logical constraints, and tone for response ranking • Produced SFT training data text pairs for dense domain coverage • Ran targeted adversarial prompt simulation matrices for red-teaming alignment • Updated datasets to mitigate hallucinations and improve model alignment quality

2025 - 2026

Multimodal Data Labeler

AudioClassification

Completed multimodal data labeling tasks that included granular text classification, regional dialect audio verification, and computer vision image bounding. Logged and categorized data to ensure consistent queue production for validated training file submission. Supported accurate dataset preparation by verifying language-related audio characteristics and annotating visual regions. • Performed text classification and regional dialect audio verification • Added computer vision image bounding annotations • Conducted clean data logging and categorization for training readiness • Ensured validated file delivery ahead of operational deadlines

2025 - 2025

Data Annotation & Search Relevance Specialist

Text

Worked as a data annotation and search relevance specialist performing query relevance tagging and classification. Conducted Side-by-Side (SxS) response auditing to support ranking optimization while meeting premium quality and spam accuracy metrics. Isolated semantic bugs, syntax inconsistencies, and localized context gaps to protect downstream production-ready training datasets. • Tagged and classified search engine query relevance • Audited SxS model responses to assess ranking suitability • Maintained strict adherence to dataset taxonomies and client rules • Debugged semantic and context issues before dataset usage

2025 - 2025

Education

Work History