Sr. Data Scientist | VuNet Systems                                                                                                                                                                                                               May 2023 – Present | Chennai, Tamil Nadu

• Designed and developed VED Copilot, an advanced GenAI-powered chatbot leveraging agentic workflows with multi-agent collaboration to extract actionable insights from observability data via natural language, enabling seamless exploration of metrics, dashboards, and enterprise knowledge bases.

• Developed an in-house LLM-powered text-to-SQL pipeline within the VED Copilot ecosystem, allowing users to ask questions in natural language, auto-generate SQL queries, and retrieve results.

• Developed a customized RAG tool within the VED ecosystem to retrieve relevant vectorized unstructured data using semantic search and context-based similarity on indexed docs and knowledge bases.

Tools Used : OpenAI, Streamlit, LlamaIndex Code Interpreter, ClickHouse, Python

Skills : Agentic workflows, LLMOps, RAG (Retrieval-Augmented Generation).

Data Scientist | VuNet Systems       Apr 2020 – Apr 2023 | Chennai, Tamil Nadu

Workflow automation & LLM based Applications

• Developed autonomous LLM-based agents for natural language queries, delivering relevant insights, summaries, and charts to integrate workflow automation into the observability platform.

• Developed an optimized RAG based Q & A chatbot with LLama 2 to identify root causes and resolutions of incidents and alerts using historical RCA data as vectorized knowledge bases.

• Fine-tuned open-source LLMs on Hugging Face, using LangChain for prompt engineering and optimization. 

Log Analytics

• Developed a customized Bi-LSTM-based joint model for log type classification and parsing, combining intent classification and named entity recognition with an Encoder-Decoder seq2seq architecture.

• Deployed Active Learning and sampling strategies to reduce labeling effort and training time for NER-based log parsing models.

• Developed the full algorithm and source code for the VulogDNA product, covering the clustering, template merging, auto-grok(regex) pattern generation, and log parsing modules.

• Developed a novel log quality scoring mechanism that generates an aggregate score based on relevant metrics for event logs.

Time Series Analysis

• Developed a mathematical formulation and source code for the Current Health Index (CHI), a novel scoring mechanism to quantify the health of a time-series signal at the current timestamp.

• Developed a novel Bayesian percentile baselining model for anomaly detection in metric time-series signals of IT system components.

• Developed a model to determine dynamic thresholds for IT system metric time-series using Bayesian percentile and prediction error methods with interquartile range.

• Implemented a DeepAR encoder-decoder LSTM model for forecasting time-series and built an anomaly detection model using DeepAR, leveraging the Gaussian nature of the forecasting errors distribution.

• Played a leading role in developing the Operational Predictive Index (OPI), a scoring mechanism to quantify future performance of time-series signals or touchpoints of an IT system( US Patent granted)

• Developed ISP Performance Index to compare internet service quality across providers and states, delivering detailed performance insights to customers.

Tools Used : OpenAI, PyTorch, Keras-TensorFlow, HuggingFace, LangChain, DeepAR, Python

Skills : LLMOps, Deep Learning, Natural Langauge Processing, LoRA & QLoRA, Hierarchical clustering, Time Series Analysis(Forecasting, anomaly detection etc), Log Analytics

Sr. Knowledge Expert (Mechanical) | T.I.M.E. Education Pvt. Ltd.   Jul 2016 – Dec 2019 | New Delhi

• Managed classroom training for the Mechanical Engineering GATE batch, a national-level entrance exam for undergraduate engineering students.

• Taught the courses(Strength of Materials, Fluid Mechanics, Theory of Machines, Applied Thermodynamics) and delivered technical seminars and workshops at various engineering colleges nationwide.

Assistant Manager | Escorts Kubota Limited Aug 2011 – Oct 2014 | Faridabad, Haryana

• Product Development & Quality Assurance: Led product development (Gears & shafts) in collaboration with Process and R&D teams and supervised quality control for gear and shaft manufacturing, including in-process and pre-dispatch inspections.

• Learning & Development: Core team member responsible for designing learning modules for 700+ blue-collar employees and delivering 35+ technical workshops at Escorts.

• Rishi Mishra,  Smriti, Ganapathy Krishnamurthi, Balaji Srinivasan, Sundararajan Natarajan, Eig-PIELM: an efficient mesh-free method for eigenvalue problems using physics-informed extreme learning machines,Computer Methods in Applied Mechanics and Engineering, Volume 451, 2026, 118674, ISSN 0045-7825, https://doi.org/10.1016/j.cma.2025.118674. (https://www.sciencedirect.com/science/article/pii/S0045782525009466)

• Influence of Thermal Boundary Conditions on Initiation of Compressor Mode in Steam Turbines, Rishi Mishra, Sawan S. Sinha and Shrish Shukla, 6th International and 43rd National Conference on Fluid Mechanics and Fluid Power, December 2016 (Paper ID 184).

• Numerical Investigation of Flow Through Steam Turbines using The Mixing Plane Method, Shrish Shukla, Rishi Mishra, Nikhil Chaudhary and Sawan Suman Sinha, International Congress on Computational Mechanics and Simulation (ICCMS), IIT Bombay, June 2016, (Ref. ID 815).

• "Top Performer" Award at VuNet Systems– Recognized developing the vuLogDNA product for log data parsing and analytics, the first ML module, through cross-functional collaboration and agile sprints under tight deadlines.

• "Excellent faculty of the year" Award at T.I.M.E. Education– Awarded consecutively for 3 years as excellent faculty of the year.

• Secured 1st position in college and 3rd rank statewide during my Diploma studies.

• Won various awards across multiple forums and platforms for classical Indian singing.