Hello, I'm

Chakshu Dhannawat.

AI Engineer.

Building intelligent systems — from RAG pipelines and multi-agent orchestration to fine-tuned LLMs. Currently at Otsuka Corporation, Tokyo.

About

I'm an AI Engineer at Otsuka Corporation in Tokyo, where I design and deploy production LLM systems — RAG chatbots, multi-agent pipelines, and fine-tuned models serving thousands of users.

I graduated from the Indian Institute of Technology, Jodhpur with a B.Tech in Artificial Intelligence & Data Science (Department Rank 2), specializing in Computer Vision. My work spans the full ML lifecycle: from data curation and model training to evaluation and deployment.

Beyond engineering, I'm a FIDE-rated chess player (1551) and hold a Japanese language certification (NAT-TEST 5Q).

            Rank 2
            Department Rank
IIT Jodhpur, AI&DS
          

            9.25
            Specialization GPA
Computer Vision
          

            10+
            LLMs fine-tuned
SFT, RLHF, LoRA
          

            1000+
            Monthly users
RAG systems served
          

Experience

AI Engineer · Otsuka Corporation Oct 2024 — Present

Tokyo, Japan

Built and deployed RAG chatbots on the ELK stack, improving retrieval accuracy by 31%, cutting search time by ~23%, and serving 1,000+ monthly users.
Scaled quotation/FAQ assistants with vector + hybrid search and light multi-agent orchestration (memory + planning); automated 60% of routine queries.
Executed LLM fine-tuning (SFT, instruction tuning, RLHF) on 10+ LLMs using LoRA/QLoRA across multi-node on-prem GPUs; validated with standardized benchmarks.

Data Scientist · Laika AI Jun — Sep 2024

Remote

Architected a multimodal LLM chatbot (text + on-demand charts) and custom blockchain tools using AutoGen; improved task completion by ~13%.
Refined data access patterns and caching with Redis, cutting external API calls by ~20%.

AI Engineer Intern · Otsuka Corporation Jun — Jul 2023

Tokyo, Japan

Piloted an LLM agent simulation (movement, planning, interaction, memory) with a 5-member team, increasing long-horizon task completion.
Designed an evaluation pipeline scoring plans and actions on success, safety, and latency with tracing and replay.

Skills

Programming

PythonCC++BashSQL

LLMs & Agents

LangChainLlamaIndexMCPAutoGenCrewAIHugging FacevLLMLoRA / QLoRARLHFRAG

ML / Deep Learning

PyTorchTensorFlowscikit-learnOpenCVNumPyPandasWeights & Biases

Cloud & MLOps

AWSAmazon BedrockAmazon SageMakerGCPVertex AIAzureAzure AI FoundryMLflowLangFuseDeepEval

Databases & Search

ElasticsearchPostgreSQLMongoDBRedisNeo4jMySQLFAISSChromaDB

Infrastructure

DockerKubernetesGitLinuxCI/CDNginx