Srashti Goyal

Srashti Goyal – Portfolio

Astrophysics and machine learning researcher with a PhD in physics and 8 years of experience. I build deep learning, generative AI and Bayesian inference systems for large-scale time-series data, and applied AI tools with LLMs, RAG and agents.

Skills

Generative AI & NLP
LLMs, NLP, RAG (hybrid, SQL), AI agents, LangChain, LangGraph, normalizing flows
Machine learning
PyTorch, TensorFlow, scikit-learn, CNNs, deep learning, feature engineering
Statistics
Bayesian inference, MCMC, hypothesis testing, time-series forecasting
Programming
Python, SQL, C, MATLAB
Deployment & tools
FastAPI, Streamlit, Docker, Qdrant, AWS Lambda, GitLab CI/CD, CUDA, Google Earth Engine

Employment

Astrophysics & ML Researcher

IUCAA, Pune, India

  • Advancing gravitational-wave lensing searches by applying machine learning and Bayesian inference to data from the global detector network.

Python · Machine learning · Bayesian inference · Time-series analysis

Researcher & Data Scientist

Max Planck Institute for Gravitational Physics, Germany

  • Cut parameter-estimation compute 10x by training and deploying normalizing-flow generative models in PyTorch on GPUs. Code · Paper
  • Improved lensed-signal detection accuracy by ~80% by designing a novel Bayesian detection statistic. Code · Paper
  • Sped up wave-optics signal modeling 100x over state-of-the-art methods with numerical integration and optimized Fourier transforms in the open-source package GLoW. Code · Paper

Python · PyTorch · Generative AI · CUDA · Bayesian inference · Distributed computing

PhD Scholar & Data Scientist

ICTS-TIFR, Bengaluru, India

  • Cut lensing search latency 1000x on 5,000+ astrophysical events by building LensID, an end-to-end CNN pipeline. Code · Paper
  • Enabled lensing-based tests of general relativity by developing scalable Bayesian hypothesis-testing frameworks. Paper
  • Forecast COVID-19 trajectories with ensemble SEIR models, applying the same statistical toolkit to epidemiology. Code · Paper

Python · TensorFlow · CNNs · Deep learning · MCMC · Time-series forecasting

Education & Certifications

PhD, Physics

ICTS-TIFR, Bengaluru, India

BS-MS Dual Degree, Physics

IISER Kolkata, India

Certified AI Engineer

Codebasics · Certificate

Achievements

Co-curricular

Wellbeing & sustainability
Certified yoga instructor, Hatha & Vinyasa (Asha Yoga Center, 2023); Permaculture Design Course (Aranya Farm, 2026)
Teaching & volunteering
Teaching volunteer, Ek Pehal initiative, IISER Kolkata (2016); volunteer scientific advisor, TANIT New Earth Solutions, Berlin (2024)
Science outreach
LIGO-India public outreach for schools and the public; speaker at the Young Astronomers' Meet (YAM), ARIES; IISc Open Science Day
Hobbies
Dancing and hiking

Projects on GitHub

LandIntel · AI land intelligence for Madhya Pradesh

Fuses 8 geospatial data sources in a fault-tolerant pipeline; an LLM agent with tool routing evaluates site suitability and powers grounded Q&A, reports and ranked site comparisons. Capstone, Codebasics AI Engineering.

Python · LLM agents · Google Earth Engine · Streamlit · Groq  |  Live app · GitHub

MediBot · Role-aware hospital assistant

Role-based access control enforced at retrieval time, hybrid dense + BM25 search with cross-encoder reranking, and SQL RAG for analytical questions; FastAPI backend and Next.js frontend with source citations.

Python · LangChain · Qdrant · FastAPI · Next.js  |  GitHub

LensIDCNN pipeline that finds strongly lensed gravitational-wave pairs
Dingo (lensing)Normalizing-flow inference for microlensed gravitational waves
GLoWWave-optics lensing package (co-developer)
PO 2.0Fast Bayesian search for strongly lensed gravitational waves
glow_waveformsLensed waveforms for compact binaries, linked to Bilby and PyCBC
COVID-19 IndiaEnsemble SEIR models and analysis of the pandemic in India

All repositories on GitHub →