Nakshatra Singh Raghav profile photo 1 on Tal

Nakshatra Singh Raghav

Backend & Applied AI Engineer @ Lucio AI | RAG, AI Agents, Distributed Systems | AWS Certified Solutions Architect

Bengaluru

3+ years of work experience

Lucio AI
Sep 20241 yr 10 mosPresent

Bengaluru, Karnataka, India

1. Legal AI Workspace: Defined and delivered a collaborative AI workspace for long-running legal matters, turning fragmented chats, documents, and research into shared matter memory for legal teams while improving prefiltering latency by 5x and designing for 100K+ document matters. 2. Legal Document Translation: Built a layout-preserving document translation pipeline adopted by enterprise customers, contributing to a ~$42K/year contract. 3. GDPR Deletion Infrastructure: Designed deletion workflows across MongoDB, blob storage, generated artifacts, and vectors, enabling post-engagement cleanup and safe user or organization offboarding across multi-million-record tenants. 4. Model Management Stack: Built multi-environment, multi-provider LLM platform tooling for model upgrades, deprecations, global overrides, on-prem setups, and environment-specific availability, reducing manual coordination for model operations. 5. Long-Running Assistant Memory: Improved instruction following in long AI sessions by designing proactive context compaction that summarized older history before limits were hit while preserving recent turns verbatim. 6. Document Processing Reliability: Improved reliability of a critical document-processing service by replacing memory-leaky processing libraries and shifting compute from VM workers to single-worker Celery containers, so one crashing worker can no longer affect the rest. 7. Legal Search Agent: Built a natural-language search agent for RSGI, a legal data platform that answered questions about lawyers, firms, ratings, and news with citations back to source pages and validating outputs against a golden dataset. 8. Agent Framework Patch: Authored the AutoGen patch for Gemini 3 compatibility issues, fixing thought- signature handling that broke multi-turn agent calls and enabling a later migration from costly Sonnet workloads back to Gemini models.
HashedTokens
Jun 20244 mosSep 2024

Remote

ACM-VIT CHAPTER
Mar 20221 yr 1 moMar 2023

• Engaged as a core committee member in ACM VIT, making substantial contributions to the planning and execution of various technical events and initiatives. • Actively supervised a project during Hacktoberfest, orchestrating more than 10 pull requests and fostering seamless collaboration among contributors. • Collaborated with team members to strategize and execute the development of the ACM Recruitment Website, Reverse Coding Portal, and Cryptic Hunt projects.

Education

Vellore Institute of Technology

Bachelor of Technology - BTech, Computer Science

Vellore Institute of Technology

20214 yrs2025
salwan public school

High School Diploma

salwan public school

Join Nakshatra and 10k+ other professionals on

Tal Boss

Tal is the app where bosses meet engineers & builders directly, without middlemen

Download Tal Boss

Nakshatra Singh Raghav on Tal

Headline

Backend & Applied AI Engineer @ Lucio AI | RAG, AI Agents, Distributed Systems | AWS Certified Solutions Architect

About

Backend & Applied AI engineer with ~2 years of experience building production AI systems for legal teams. At Lucio AI, I’ve worked across agents, retrieval systems, document processing, translation pipelines, model management, and backend reliability. My work has ranged from reading syslogs of services OOMing in production and fixing document-processing failures, to designing agent orchestration, retrieval flows, long-running assistant memory, and GDPR-safe deletion infrastructure. I like working on systems where correctness, trust, and reliability matter: RAG, AI agents, search, backend infrastructure, observability, and distributed systems. Open to Backend Engineering and Applied AI Engineering roles nakshatra.raghav@gmail.com

Location

Bengaluru

Work experience

Software Development Engineer at Lucio AI, Sep 2024 - Present · 1 yr 10 mos, Bengaluru, Karnataka, India, 1. Legal AI Workspace: Defined and delivered a collaborative AI workspace for long-running legal matters, turning fragmented chats, documents, and research into shared matter memory for legal teams while improving prefiltering latency by 5x and designing for 100K+ document matters. 2. Legal Document Translation: Built a layout-preserving document translation pipeline adopted by enterprise customers, contributing to a ~$42K/year contract. 3. GDPR Deletion Infrastructure: Designed deletion workflows across MongoDB, blob storage, generated artifacts, and vectors, enabling post-engagement cleanup and safe user or organization offboarding across multi-million-record tenants. 4. Model Management Stack: Built multi-environment, multi-provider LLM platform tooling for model upgrades, deprecations, global overrides, on-prem setups, and environment-specific availability, reducing manual coordination for model operations. 5. Long-Running Assistant Memory: Improved instruction following in long AI sessions by designing proactive context compaction that summarized older history before limits were hit while preserving recent turns verbatim. 6. Document Processing Reliability: Improved reliability of a critical document-processing service by replacing memory-leaky processing libraries and shifting compute from VM workers to single-worker Celery containers, so one crashing worker can no longer affect the rest. 7. Legal Search Agent: Built a natural-language search agent for RSGI, a legal data platform that answered questions about lawyers, firms, ratings, and news with citations back to source pages and validating outputs against a golden dataset. 8. Agent Framework Patch: Authored the AutoGen patch for Gemini 3 compatibility issues, fixing thought- signature handling that broke multi-turn agent calls and enabling a later migration from costly Sonnet workloads back to Gemini models.

Software Development Engineer at HashedTokens, Jun 2024 - Sep 2024 · 4 mos, Remote

Core Committee Member at ACM-VIT CHAPTER, Mar 2022 - Mar 2023 · 1 yr 1 mo, • Engaged as a core committee member in ACM VIT, making substantial contributions to the planning and execution of various technical events and initiatives. • Actively supervised a project during Hacktoberfest, orchestrating more than 10 pull requests and fostering seamless collaboration among contributors. • Collaborated with team members to strategize and execute the development of the ACM Recruitment Website, Reverse Coding Portal, and Cryptic Hunt projects.

Education

Bachelor of Technology - BTech, Computer Science, Vellore Institute of Technology

High School Diploma, salwan public school