Satya Saurabh Mishra — Data Scientist

I am a Data Scientist at Dell Technologies with expertise in Text-to-SQL systems, Financial AI, and Generative AI. My work lies at the intersection of Information Retrieval, LLMs, and Applied Machine Learning, with a strong focus on building agentic frameworks, RAG-based systems, and AI-powered automation that drive measurable business outcomes. At Dell, I have: - Published ICLR 2025 research on entropy-guided Text-to-SQL, improving confidence estimation and cutting redundant generations. - Automated 6,000+ quarterly balance sheet flux analyses using a custom LLM pipeline, saving 32,000+ manual hours annually. - Built agent-driven workflows for retrieval, query refinement, and auto-commentary generation in finance analytics. - Enhanced enterprise search through fine-tuned embeddings, vector databases (FAISS, Chroma, PGVector), and optimized RAG pipelines. - My broader experience includes prompt engineering, SQL evaluation strategies, LLM fine-tuning, and MLOps practices (MLflow, Airflow, DVC, Git, Docker, K8s). I’m particularly passionate about pushing the limits of AI Agents for enterprise-scale adoption and designing end-to-end ML solutions that combine robustness with innovation. Beyond hands-on implementation, I am deeply interested in applied research and open-source contributions. we have released finance-specific LLM and embedding models trained on 50M-token datasets, and continue to explore ways to make AI systems more reliable, interpretable, and impactful. Recognized with Dell’s Game Changer Award for AI-driven innovation, I am motivated to keep learning, mentoring, and building systems that transform data into meaningful intelligence I believe in learning deeply, building boldly, and sharing generously. Let’s connect!

Stackforce AI infers this person is a Data Scientist specializing in AI and Machine Learning for Fintech and enterprise solutions.

Location: Bengaluru, Karnataka, India

Experience: 5 yrs 8 mos

Skills

Applied Machine Learning
Information Retrieval
Ai-powered Automation
Natural Language Processing
Machine Learning
Computer Science

Career Highlights

Automated 6,000+ quarterly analyses, saving 32,000+ hours annually.
Published research on entropy-guided Text-to-SQL systems.
Recognized with Dell’s Game Changer Award for AI-driven innovation.

Work Experience

Dell Technologies

Data Scientist (1 yr 10 mos)

Data Science Intern (10 mos)

Indian Institute of Information Technology, Design and Manufacturing, Jabalpur

Teaching Assistant (1 yr 9 mos)

SSV academy Jaunpur

Mathematics Teacher (2 yrs 1 mo)

Education

Master of Technology - MTech at Indian Institute of Information Technology, Design and Manufacturing, Jabalpur

Bachelor of Technology - BTech at Veer Bahadur Singh Purvanchal University (VBSPU)

Satya Saurabh Mishra

Data Scientist

Bengaluru, Karnataka, India5 yrs 8 mos experience

AI EnabledAI ML Practitioner

Key Highlights

Automated 6,000+ quarterly analyses, saving 32,000+ hours annually.
Published research on entropy-guided Text-to-SQL systems.
Recognized with Dell’s Game Changer Award for AI-driven innovation.

Stackforce AI infers this person is a Data Scientist specializing in AI and Machine Learning for Fintech and enterprise solutions.

Contact

Skills

Core Skills

Applied Machine LearningInformation RetrievalAi-powered AutomationNatural Language ProcessingMachine LearningComputer Science

Other Skills

Text-to-SQLEntropy-guided refinementAgentic frameworksLLM pipelinePrompt engineeringData preprocessingVariance analysisRAG methodologyLLM fine-tuningVector databasesPrompt techniquesMLflowGitlabDockerKubernetes

About

Experience

5 yrs 8 mos

Total Experience

1 yr 11 mos

Average Tenure

1 yr 10 mos

Current Experience

Dell technologies

2 roles

Data Scientist

Jul 2024 – Present · 1 yr 10 mos · Bengaluru, Karnataka, India · On-site

ICLR 2025 Accepted Research Paper: How Does Entropy Influence Modern Text-to-SQL Systems?
Proposed an entropy-based metric to assess and cluster SQL query candidates for improved confidence
estimation.
Integrated entropy-guided refinement into CHESS and CHASE pipelines to reduce redundant generations.
Built a modular agentic framework using Mixtral and Deepseek models for retrieval, generation, and selection.
Applied MarkupLM with DBSCAN to cluster execution-based embeddings of SQL outputs.
Used entropy thresholds as stopping criteria in query refinement to cut down computation costs.
Ran detailed experiments on BIRD benchmark using diverse generation strategies like Divide-and-Conquer, Query
Plan, Online Synthetic Examples.
AI-Powered Balance Sheet Flux Commentary Automation
Automated 6,000 quarterly flux analyses using a custom-built LLM pipeline, reducing 32K+ hours of manual
effort annually.
Conducted extensive prompt engineering experiments and adopted Chain-of-Thought prompting for best results.
Fine-tuned a Llama-3.1-8B-Instruct using PEFT with LORA on historical finance data to improve accuracy.
Designed and implemented an tool based agentic framework to orchestrate data preprocessing, variance analysis, and
auto-commentary generation.

Text-to-SQLInformation RetrievalApplied Machine LearningEntropy-guided refinementAgentic frameworksAI-powered automation

Data Science Intern

Jul 2023 – May 2024 · 10 mos · Bengaluru, Karnataka, India · On-site

Intelligence Search
Enhanced LLM model’s query response using RAG methodology with Dell data.
Conducted end-to-end experiments on llama-index and langchain, integrating OpenAI and custom LLM models.
Led efforts to fine-tune embedding models, improving their ability for enhanced dell data representation.
Explored Vector Database like PGVector,Chroma DB and Faiss for efficient storage and retrieval.
Implemented Different prompt techniques like Chain of Thought, Tree of Thought, meta prompting etc. for optimizing and getting better results from LLM.
optimized model performance through different evaluations strategy using Prometheus, Deep Evaluation, RAGAS etc. and comparisons with other LLM models.
Integrated the MLflow (Model Versioning) tool for analyzing and monitoring the performance of AI models.
Used Gitlab (code versioning) , DVC (Data versioning) , Docker, K8s etc. for deployment.