Skip to content
View DeepanJayaraman's full-sized avatar

Block or report DeepanJayaraman

Block user

Prevent this user from interacting with your repositories and sending you notifications. Learn more about blocking users.

You must be logged in to block users.

Content in all repositories owned by your account will be closed.
Maximum 250 characters. Please don’t include any personal information such as legal names or email addresses. Markdown is supported. This note will only be visible to you.
Report abuse

Contact GitHub support about this user’s behavior. Learn more about reporting abuse.

Report abuse
DeepanJayaraman/README.md

Deepan Jayaraman

Ph.D., IIT Madras  ·  Data Science, Caterpillar India

Deciding well when the data is scarce and the tails matter.

Website Google Scholar ORCID LinkedIn


Welcome, and thanks for stopping by. I'm a quantitative researcher and a working data scientist, and everything here comes out of one question.

Data-driven decision making

How choices should be made when the evidence is thin.

  • The problem — consequences are asymmetric, and the model informing the choice is itself estimated rather than specified.
  • The methods — uncertainty quantification, probabilistic risk assessment, robust and multi-objective optimisation.
  • They travel — the same question turns up in business planning and in engineering reliability.

Applications

Business Operational risk · Demand planning · Resource allocation · Decision support · Forecasting
Engineering Reliability · Fatigue life · Robust design · Motor design · Tail risk · Surrogate modelling

The code behind it

Toolboxes, research studies and the analyses behind the papers. The writing lives on my site — see below.

Methods and software   LMomFit fitting · L_UQ uncertainty · RDO robust design · PRA risk assessment

Research studies   lmrd-selection-bias · lmrd-set-coverage · lmoment-pot-reliability · conformal-dro-aviation · asrs-atc-detection-gaps · predict-explain-prescribe · ai-driven-product-design

Applied work   fatigue-life-screening · CPM_prediction · Timeseries · Optimization · Visuvalization · DeepanJayaraman.github.io — this site

Python R MATLAB scikit-learn XGBoost TensorFlow LangChain Ollama SQL Power BI Azure

On the website

Research The doctoral contribution and the current threads, with the methods drawn out
Publications Journal articles, manuscripts under review, conferences and invited talks — filterable by type, domain and rating
Industry Caterpillar, Onward Technologies and Siemens, with the stack and application areas per project
Teaching Courses taught and courses I can offer

Pinned Loading

  1. A-Probabilistic-sample-level-ML-based-Safety-Screening-Framework-for-Fatigue-Life A-Probabilistic-sample-level-ML-based-Safety-Screening-Framework-for-Fatigue-Life Public

    MATLAB framework for probabilistic S-N curve generation and ML-based fatigue life classification, using L-moments for parameter estimation from small samples.

    MATLAB

  2. ai-driven-product-design ai-driven-product-design Public

    Reproducibility package for "From Intuition to Intelligence": a simulation-based framework coupling predictive ML, explainable AI, and multi-objective optimization for early-stage product design.

    Python

  3. asrs-atc-detection-gaps asrs-atc-detection-gaps Public

    Reproduction code for 'Detection gaps in the pilot-controller system': a four-decade BERTopic and detection-pathway analysis of 30,410 ATC-related NASA ASRS incident reports (1988-2026). Submitted …

    Python

  4. conformal-dro-aviation conformal-dro-aviation Public

    Coverage-calibrated ambiguity sets for distributionally robust sourcing under certification-constrained substitution, evaluated on FAA Service Difficulty Report data.

    Python

  5. LMomFit LMomFit Public

    L-moment distribution fitting for small samples containing extremes. Parallel MATLAB and Python, with guarded automatic fitting and bootstrap family selection.

    Python

  6. predict-explain-prescribe predict-explain-prescribe Public

    Decision-support artifact in which explanation gates prescriptive search: calibrated simulation, DP2 attribution gate, and Pareto frontier over a learned objective. Replication code and data.

    Python