Sakshi Ray

AI Researcher

Bengaluru, Karnataka, India1 yr 5 mos experience

AI EnabledAI ML Practitioner

Key Highlights

Expert in building scalable GenAI applications.
Proficient in LLMOps and security guardrails.
Experience with multimodal RAG solutions.

Stackforce AI infers this person is a Generative AI Engineer specializing in LLM-driven systems for enterprise applications.

Contact

Skills

Core Skills

Generative AiLlmops

Other Skills

Large Language Models (LLM)Large Language Model Operations (LLMOps)FastAPIDjangoDockerKubernetesAzureCybersecurityPython (Programming Language)Microsoft Power BIMicrosoft ExcelClaude SkillsModel Context Protocol (MCP)AgentsAmazon Web Services (AWS)

About

AI Engineer specializing in Generative AI and LLM‑driven systems, focused on building scalable, production‑ready GenAI applications. Experienced in LLMOps, prompt engineering, and LLM evaluation, with hands‑on expertise using FastAPI, Django, Docker, Kubernetes, and Azure, combined with vector databases, and cloud‑native infrastructure, to deliver secure, observable, and production‑ready AI applications. I specialize in Generative AI and agentic architectures, building production‑grade RAG and multimodal RAG solutions that enable intelligent retrieval, grounding, and reasoning over enterprise‑scale knowledge bases. I have designed and implemented multi‑agent systems leveraging agent routing, planning, shared memory, and orchestration patterns to solve complex, multi‑step workflows with clear execution boundaries. I also work extensively on security and guardrails for GenAI applications, including prompt‑injection mitigation, content filtering, data‑leakage prevention, and controlled tool execution, ensuring compliant, trustworthy chatbot deployments in real‑world environments. My experience includes building synthetic data generation pipelines and integrating autonomous agents using MCP‑based frameworks, enabling structured multi‑step reasoning, controlled tool execution, and reliable decision‑making workflows in production AI systems.

Experience

1 yr 5 mos

Total Experience

1 yr 5 mos

Average Tenure

1 yr 5 mos

Current Experience

Fractal

2 roles

Associate - AI Engineer

Nov 2024 – Present · 1 yr 5 mos

Designed and implemented an end‑to‑end synthetic data generation pipeline to produce high‑quality, domain‑specific datasets for LLM fine‑tuning and evaluation, leveraging both Chain‑of‑Thought and non‑CoT prompting strategies.
Built a production‑grade multimodal RAG system enabling text‑image understanding, incorporating metadata‑driven retrieval, smart re‑indexing strategies to reduce embedding costs, and evaluation‑driven tuning to achieve high‑precision, grounded responses.
Designed and implemented secure RAG agents with identity‑aware document access, embedding‑level safeguards, and role‑based authorization, preventing unauthorized data exposure in enterprise environments.
Implemented comprehensive LLM security guardrails, including prompt‑injection detection, toxicity and unsafe content filtering, PII masking, token‑abuse controls, and controlled tool execution, ensuring trustworthy GenAI deployments.
Developed multi‑agent LLM orchestration frameworks, implementing supervisor and specialized data agents for intelligent query routing, coordinated multi‑step reasoning, and reliable agent execution with clear control boundaries.
Built and integrated Microsoft Teams chatbots with Adaptive Cards, enabling rich, interactive user experiences for agent‑driven workflows. Designed conversational flows that allow users to interact with deployed AI agents, visualize responses, trigger actions, and receive contextual insights directly within Teams.
Developed backend services with FastAPI and Django for RAG pipelines, and agent‑based execution layers.
Defined and executed LLM evaluation frameworks, leveraging metrics such as BLEU, METEOR, ROUGE, Perplexity, and task‑specific evaluations, continuously monitoring response quality, grounding, and safety.
Deployed and operated GenAI systems on Azure Kubernetes Service (AKS) using Docker, Helm, and automated CI/CD pipelines, ensuring scalability, reliability of LLM workflows.