Headline

Lead AI/ML Scientist | Decision Systems, Optimization & Execution Governance | MSc Artificial Intelligence (UK)

About

I’m a Lead AI/ML Scientist working on large-scale, production-grade AI systems across optimization, analytics, and decision support in complex enterprise environments. Over the last 10+ years, I’ve been involved end-to-end in AI initiatives — from problem framing and data readiness to model development and operational adoption — particularly in settings where business complexity, scale, and execution constraints matter as much as technical accuracy. A recurring pattern I’ve observed is that successful AI adoption depends less on model sophistication and more on execution clarity: clear ownership, disciplined scoping, architectural judgment, and realistic expectations about what should (and should not) be built. Alongside my core role, I spend time engaging with technical leaders and early-stage founders on use-case validation and execution design — helping teams structure pilots, stress-test assumptions, and avoid common delivery pitfalls before scaling. My areas of focus include: Mathematical Optimization & Operations Research – scheduling, routing, and resource allocation problems Predictive Analytics at Scale – decision-oriented forecasting beyond dashboards Generative AI & Knowledge Systems – RAG and agent-based systems for internal decision support Enterprise Digital Twins – simulation of complex operational scenarios I value thoughtful, grounded conversations around practical AI adoption and decision systems in real-world environments. Always open to connecting with peers who are navigating complex execution challenges.

Location

Bengaluru

AI stack

anthropic apis

claude code

claude

github copilot

hugging face

langgraph

Work experience

Lead Data Scientist at A.P. Moller - Maersk, Mar 2021 - Present · 5 yrs 9 mos, Bengaluru, Karnataka, India, Lead Data Scientist for Enterprise Performance Management System, driving profitability via AI/ML implementation. Pioneered Gen-AI based Autonomous Agents: Developed Insight Agent for contextual tabular data insights, self-healing Pandas/SQL agents, Hierarchical Multi-Agent systems, and a MemGPT-inspired Multi-Agentic Memory system for enhanced context and learning. Optimized LLM Deployment: Finetuned SLMs (QLoRA) and deployed on GPT-enabled VMs using vLLM/TensorRT engines with dynamic batching, reducing inference time by ~80%. Advanced RAG Implementation: Engineered sophisticated RAG techniques, including intent-based search, dynamic indexing, ranking, and Hybrid RAG with knowledge graphs. Core ML Achievements: Delivered revenue forecasts with >95% accuracy and developed a SHAPley-based Root Cause Analysis framework for data-driven business actions. Built Digital Twins of port terminals using Simpy for operational efficiency. Strategic Integration: Building frameworks to integrate Performance Management Systems with Autonomous Agents and finetuned LLMs

Data Scientist at MyYogaTeacher, Oct 2019 - Feb 2021 · 1 yr 5 mos, Bengaluru Area, India, Lead Data Science team to acquire , track & analyze data and also build machine learning and deep learning models to improve data understanding and user experience Yoga Pose Recognition: Built Automatic live yoga pose recognition using TensorPose and extensive OpenCV concepts with deep CNN's in play Bad Users Out: Built an intelligent system to recognize and detect foul users out of the yoga sessions using face recognition and automatic training techniques Session summary : Created summarizer to generate automatic abstract summary from session transcripts using BERTSUMM Data integrator : Integration of data from facebook ads, Google analytics ,click stream analytics and MYSQL to AWS redshift which is in turn connected to Tableau A/B master : Built A/B testing framework using advance statistical techniques like statistical significance, effect size and confidence intervals to derive the results Predictors : Extensive use of machine learning models like Naive Bayes, SVM, Logistic regression, XGBoost

Data Analyst at NTT DATA, Sep 2016 - Oct 2019 · 3 yrs 2 mos, Bengaluru, Karnataka, India, Production Failure Predictor : Used SVM and Boosting techniques to predict production job failures there by significantly reducing critical issue and system down time Data Traffic :Built and deployed time series model using Holt Winters to predict data traffic to manage overall load on infrastructure SQL development :Developed database objects, including tables, views and materialized views using SQL.

Software Engineer at NTT DATA, Dec 2015 - Sep 2016 · 10 mos, Bengaluru Area, India, For WELLS FARGO Responsible for understanding the clientele requirements and calibrate the same with business in building Machine Learning algorithms and software applications . -Worked on Analyzing the data to find the data pattern causing business issues using Python -Built ETL packages using Python/SSIS to process large data -Built Predictive ML modules to avoid business issues - Customer satisfaction for application enhancements on the current platform - Customer interactions with cross country collaborations - Responsible for training the new resources on project applications - Constantly working on the application optimization Application platforms includes: DotNet/.Net & SQL

Co-Founder & CTO at SevaToday - India, Aug 2011 - Oct 2019 · 8 yrs 3 mos, Bengaluru Area, India, SevaToday is an online platform catering to the needs for various festive occasions. It will help the customers to pre-book the pundits for religious activities. - Responsible for building the entire software architecture - Developed the platform end to end using WordPress - Constant bug fixing and enhancements

Education

Certification, Quantum Computing and Machine Learning, Indian Institute of Technology, Delhi

Master of Science - MS, Machine Learning & Artificial Intelligence, Liverpool John Moores University

Public links

github: https://github.com/NandaKishoreJoshi