Saurabh Sagar Shinde
Researcher in uncertainty-aware decision intelligence
I study how machine-learning predictions should be trusted when real decisions depend on them.
I’m a computer engineering undergraduate at K. K. Wagh Institute of Engineering Education & Research in Nashik, working on uncertainty-aware decision intelligence: predicting how long a machine will keep working, attaching honest uncertainty to that prediction, and turning both into a maintenance decision.
Alongside research I build systems people use — most visibly Rakshak AI with Nashik Rural Police. Seeing how deployed systems are actually used, and where they fail, shapes which research questions I think are worth asking.
- Uncertainty-aware decision intelligence
- Decision-theoretic predictive maintenance (NASA C-MAPSS)
- 1 manuscript in preparation · 2 preprints · 1 published article
- B.E., Computer Engineering · KKWIEER, Nashik · expected July 2027
- Co-developer of Rakshak AI · co-founder of Neuronex Developers
A prediction is only useful if you know how far to trust it
When a model predicts how long a machine will keep working, how much should a maintenance decision trust that prediction — and what should it do with the uncertainty?
Most predictive-maintenance work asks how accurately a model can estimate a machine’s remaining useful life (RUL). I’m interested in the step after that: a maintenance planner has to decide when to intervene, and a single number does not say how wrong it might be.
My current project connects three pieces. A deep time-series model predicts RUL; conformal prediction wraps that prediction in an interval with a coverage guarantee; and a decision layer chooses the maintenance time that minimises expected cost, trading the risk of failure against the cost of acting early.
The framework, implementation and experimental protocol are mine. The manuscript is in preparation, and no empirical results are reported on this site until the experiments are complete.
- 01
Predict
A deep time-series model estimates remaining useful life.
- 02
Quantify
Conformal prediction turns the estimate into an interval with a coverage guarantee.
- 03
Decide
An expected-loss rule picks the maintenance time, trading failure risk against acting early.
Research interests
- Predictive maintenance / prognostics and health management
- Remaining useful life prediction
- Uncertainty quantification
- Conformal prediction
- Decision-theoretic machine learning
- Risk-aware decision making
- Time-series deep learning
- Transformers and temporal convolutional networks
- Business analytics
- Intelligent maintenance systems
- AI agents and multi-agent systems
Publications and manuscripts
- Decision-Theoretic Predictive Maintenance with Calibrated Predictive Uncertainty for Remaining Useful Life Prediction
- Enterprise AI Agent Ecosystems: Architecture, Economics, and the Composition Gap
- Justice-as-a-Protocol: Designing a Stateless, Scalable, and Self-Enforcing Online Dispute Resolution (ODR) System for 21st Century India
- From Chariots to Cavalry: The Evolution of Indian Warfare and the Fall of the War Chariot
From systems people use to questions worth studying
Rakshak AI began in a cybersecurity session at K. K. Wagh, when ASP Aditya Mirkhelkar challenged students to rethink technology for public safety. Five of us took it on, guided by Prof. Dhananjay Kanade and Dr. Snehal Kamalapur. We mapped police services into citizen-friendly intents, organised the relevant procedures and laws, and built a bilingual (Marathi and English) assistant with retrieval over that material.
It launched on WhatsApp with Nashik Rural Police on 1 September 2025. A year on, the team reported more than 5,000 users and 51,376 police-related conversations. The team summed up that year as a loop: build, deploy, listen, measure, find the gaps, improve. It also exposed a gap — WhatsApp assumes a smartphone and a data connection — which is why the team is now working on access through an ordinary phone call (IVR).
Seeing how deployed systems are actually used, and where they fail, shapes which research questions I think are worth asking.
- users
- 5,000+
- police-related citizen conversations
- 51,376
- complaints and reports received, assessed and resolved
- 86
Working on systems-level questions with others
With Ankur Shinde, Niraj Bava, Aditi Deodhar and Vanya Gupta, I co-authored a survey of enterprise AI agent ecosystems (SSRN, May 2026). It argues that the gap between what agents can do and the value they deliver is a ranking problem, and proposes a Trust-Aware Ranking System (TARS) for choosing agents; my part was the analysis of discovery, selection, trust, governance and economic constraints.
Earlier, in Justice-as-a-Protocol (2025), I led the system logic and built the frontend for an online dispute-resolution design for India, working out how legal rules become platform workflow.
To strengthen the quantitative side, I completed Econometrics: Methods and Applications from Erasmus University Rotterdam: regression, time-series and panel data, instrumental variables and causal inference.
Building carefully, end to end
My engineering background is full-stack development. NyayMargadarshak, built for Advocate Chinmay Gadhe, makes India’s new criminal codes navigable for citizens and connects them with lawyers; NyayBot is a legal-guidance chatbot. From September 2025 to March 2026 I interned as a full-stack developer at DreamCare Developers, working on government-oriented web applications.
In September 2025 I co-founded Neuronex Developers, a software and AI development company, where I lead as CEO.
Milestones
Dates as recorded on the CV and LinkedIn. Recognition belongs to the teams involved.
- Jun 2025Completed Econometrics: Methods and Applications (Erasmus University Rotterdam, Coursera).
- Jul 2025Justice-as-a-Protocol shared on ResearchGate (co-author).
- 1 Sep 2025Rakshak AI launched with Nashik Rural Police; the launch was covered by the Times of India, Loksatta and the Free Press Journal.
- Sep 2025Co-founded Neuronex Developers.
- 2 Jan 2026Presented Rakshak AI to the Governor of Maharashtra, Acharya Devvrat, with the Nashik Rural Police leadership.
- Mar 2026AWS Academy Graduate — Cloud Foundations.
- May 2026Enterprise AI Agent Ecosystems posted on SSRN (co-author).
- 2026Began independent research on decision-theoretic predictive maintenance; manuscript in preparation.
- Sep 2026Rakshak AI’s first year: 5,000+ users and 51,376 conversations, as reported by the team.
- 28 Sep 2026UK Design Registration No. 6549379 granted for “Communications Equipment for Public Safety” (Rakshak AI team).
Roles and responsibilities
Co-founder and CEO
Neuronex Developers
A technology company building software, automation and AI systems.
Secretary
Rotaract Club of Nashik Nine Hills
Runs communication, records and coordination for the club’s service projects; this year’s theme is “Yugantar”.
Member, editorial team
Beyond the Horizon (college magazine)
Content creation, editing and publication.
Member
Indian Society for Technical Education (ISTE), KKWIEER
Student chapter of the national technical-education society.
Recognition and learning
- Rakshak AI received institutional and public-sector recognition for its work with Nashik Rural Police, including presentation of the project to the Governor of Maharashtra on 2 January 2026.
- UK Design Registration for “Communications Equipment for Public Safety” (Rakshak AI team), 2026.
- Tech Horizon 2025 national-level hackathon — built Shakti Link under a 48-hour constraint.
- Completed Econometrics: Methods and Applications (Erasmus University Rotterdam, via Coursera, June 2025) — regression, time-series and panel-data methods, causal inference, instrumental variables.
- AWS Academy Graduate — Cloud Foundations (Amazon Web Services, March 2026).
Outside research
Interested in artificial intelligence, geopolitics, mixed martial arts (MMA). Speaks English and French.
What collaborators say
Excerpts from recommendations written on LinkedIn by teammates.
Saurabh possesses excellent research skills, allowing him to explore complex domains, evaluate solutions critically, and translate research into practical implementations.
Niraj Bava · Worked with Saurabh on the same team · 31 Jan 2026 Saurabh is collaborative, reliable, and technically sound—making him a valuable addition to any team.
Yash Vadnere · Worked with Saurabh on the same team · 17 Mar 2026
Looking for a research home.
I’m looking for research environments where prediction, uncertainty and decision-making are studied together, and for collaborators who care about evaluating methods honestly.