- AI Systems
How models, data and tools fit together into reliable systems.
- Automation Thinking
Where automation saves effort, and where it does not.
- Intelligent Workflows
AI designed around how people learn and work.
- Scalable Architectures
Designs that keep working as use grows.
Doctoral research
PhD · upcomingArtificial Intelligence for the Computational Analysis of Cosmological Data and Ancient Astronomical Knowledge
This doctoral project asks how modern AI methods can help analyse cosmological observations and, separately, how they can help study and verify astronomical knowledge recorded in ancient texts and traditions. The work keeps a strict line between four kinds of content: scientific evidence, historical interpretation, hypothesis and tradition.
Apply machine learning to public cosmological and astronomical datasets for pattern finding and classification.
Use NLP to structure astronomical statements from ancient texts (including Sanskrit sources) into a searchable dataset.
Compare recorded claims with computed sky positions and modern data, reporting agreement and disagreement openly.
- Machine learning on open astronomy datasets
- Multilingual NLP and knowledge extraction
- Astronomical computation of historical sky positions
- Transparent evaluation and labelled evidence levels
- Status: proposed. No results are claimed yet.
- Literature review and dataset survey
- Pilot study design
- Seeking a doctoral supervisor and collaborators
Current project
Since Sep 2026AI Doubt-Solving Assistant for Hindi and English Learners
Does answer quality differ between Hindi and English questions?
Does grounding answers in course notes (RAG) reduce errors?
Do learners find grounded answers more trustworthy?
Method: 50 bilingual doubts answered with and without retrieval, scored with a rubric. Data, code and a paper will be published openly.
Dataset & code ↗Publications
Books · papers · dataS. Tekriwal. AI Model Making: Basic to Advanced, Vol. 1.
50 questions for evaluating grounded AI answers.
First paper from the doubt-solving project; listed here once submitted.
Research log
Registered ORCID iD · defined scope and three research questions · built a first 50-question bilingual dataset.
Build Your First AI Models, free to read online.
Baseline answers and an evaluation rubric.
