Addankinaveen 8.

Key Skills

Software

Mercor

Labelbox

Top Subject Matter

Multi-agent document intelligence and LLM evaluation

Financial document intelligence

LLM/VLM fine-tuning

Top Data Types

Document

Image

Computer Code Programming

Top Task Types

RLHF

Routing

Data Collection

Freelancer Overview

AI Engineer at ShrinQ Consulting Group INC (June 2025–Present). Brings 7+ years of professional experience across complex professional workflows, research, and quality-focused execution. Core strengths include Don't disclose and Other. Education includes Master of Science, Indiana University Bloomington (2025). AI-training focus includes data types such as Document and Audio and labeling workflows including Evaluation, Rating, and RLHF.

Labeling Experience

AI Engineer at ShrinQ Consulting Group INC (June 2025–Present)

Don't discloseDocument

Developed an LLM observability and validation workflow to assess and improve output reliability prior to delivery. Implemented tracing and evaluation using human-in-the-loop feedback loops to iteratively refine model behavior. Built automated checks to detect low-confidence LLM outputs and reduce risk of incorrect responses. • Used LangSmith for tracing, evaluation, and continuous improvement • Created validation pipelines for low-confidence output detection • Enabled human-in-the-loop feedback to guide improvements • Implemented RAG retrieval enhancements to improve relevance

2025 - Present

AI/ML Research Engineer at Indiana University (January 2025–August 2025)

Don't discloseDocumentData Collection

Managed large-scale labeling initiatives for a financial-document agentic RAG and model improvement workflows. Supervised annotators and delivered a high-volume labeled dataset for training and evaluation purposes. Ensured labeling quality by maintaining high accuracy on produced labels. • Managed 50+ annotators • Delivered 250K+ labeled datasets • Maintained 95% labeling accuracy • Supported downstream LLM alignment and evaluation

2025 - 2025

AI/ML Research Engineer at Indiana University (January 2025–August 2025)

OtherDocumentRLHF

Built and fine-tuned agentic RAG systems over large financial document collections to improve aligned response quality. Fine-tuned LLMs and VLMs using LoRA and multi-GPU training with RLHF and DPO alignment methods. Evaluated and security-tested model outputs, including mitigation for adversarial and prompt-injection risks. • Engineered agentic RAG with hybrid retrieval across FAISS, Qdrant, and ChromaDB • Performed LLM/VLM fine-tuning with RLHF and DPO alignment • Optimized inference with quantization and deployed with vLLM • Conducted evaluation and security testing of LLM outputs

2025 - 2025

Graduate Generative AI Engineer at Indiana University (May 2024–December 2024)

OtherAudioRouting

Developed a real-time Voice RAG pipeline combining speech-to-text, semantic retrieval, and text-to-speech. Implemented automated ingestion and vector database update pipelines to support continuous knowledge refresh for conversational use. Optimized retrieval and added security enhancements to improve response relevance in the voice experience. • Integrated speech-to-text with semantic search (Pinecone) • Implemented retrieval-optimized low-latency Voice RAG • Built continuous vector database update pipelines • Improved response relevance using retrieval and security enhancements

2024 - 2024

Education

Work History