Anika Jain
A full-stack Applied AI Engineer, from paper to production.
- Field
- Retrieval · Anomaly detection · Distillation · Multi-agent
- Stack
- Python · PyTorch · FastAPI · Docker
- Source
- github.com/anidoesdev
Projects
5 systems, end to end — retrieval, anomaly detection, model distillation, and multi-agent research. 3 deployed, all open source.
Runnel
A workflow automation platform, built from scratch
A self-hosted workflow automation platform in a 9-package TypeScript monorepo. A Vue 3 canvas drives a custom graph execution engine with a hand-written expression language and a sandboxed Code node. An AI assistant builds and repairs workflows on a draft copy through validated tools, and only a person can apply that draft to the live workflow.
MemNest
Long-term memory for AI agents
A self-hostable memory engine for AI agents. Rather than returning chunks that merely resemble a query, it extracts atomic facts from raw content and decides how each one relates to what is already known — replacing, extending or reinforcing it — and a superseded fact is kept and stays traceable rather than deleted. Recall fuses lexical and vector search, packs the result to a token budget, and returns a trace saying what was dropped and why.
Papyrus
Cited answers from scientific papers
Built a full-stack Retrieval-Augmented Generation (RAG) system for scientific literature Q&A, enabling users to ask natural-language questions and receive cited, grounded answers backed by source documents.Combines dense vector search and keyword retrieval through reciprocal rank fusion, reranks candidates with an LLM relevance scorer, and synthesizes grounded answers with inline citations.
Distill
A small model that beats frontier APIs
Fine-tuned a small language model with SFT and DPO to extract structured JSON from scientific papers - authors, methodology, datasets, findings, limitations. Quantized and served with vLLM behind a FastAPI gateway with schema-constrained decoding. Outperforms frontier APIs on accuracy while cutting cost-per-call by an order of magnitude.
Synthesis
Five agents researching together
A multi-agent system that answers complex research questions across scientific literature. Five specialized agents - planner, retriever, reader, critic, synthesizer - collaborate through structured memory and tool use, with every decision logged to a custom trajectory viewer. Includes a 50-question eval harness with LLM-as-judge scoring and a hallucination-guard pipeline.
Work Experience
- Sole engineer on a contracted MVP build for a UK children's sleep product: Next.js PWA + Supabase/Postgres, shipped ahead of a 50-family validation study.
- Designed the data model and rules engine converting nightly parent logs into one human-reviewed weekly recommendation
- Enforced consent, age gating, and row-level security (RLS) at the database layer; built export, revoke, and wipe flows for UK GDPR and Children's Code compliance
- Built the operator review console and the app's shared design-system components
- The docstring of _get_train_sampler in GRPOTrainer comment to mention num_iterations > 1 fixed the existing issue.
- Engineered and optimized machine learning architectures by applying foundational mathematical principles (linear algebra, multivariate calculus, and probability), enabling the custom tuning of loss functions and optimizers that improved model accuracy by 15% over standard baselines.
- Architected, trained, and evaluated complex deep learning models using PyTorch and TensorFlow, specifically optimizing forward and backward passes to reduce training times by 30% and maximize computational efficiency on datasets exceeding 500GB.
- Spearheaded the development of an end-to-end AI research pipeline designed to process and synthesize massive, unstructured scientific data, automating the extraction of latent patterns and reducing manual analysis time by 40%.
- Implemented advanced neural network architectures, including graph-based models, directly from research papers, mathematically verifying adjacency matrices to accurately capture non-linear relationships and boost predictive performance by 22% on highly relational data.
- Constructed robust data ingestion and feature engineering workflows, resolving bottlenecks in data preprocessing to decrease data pipeline latency by 45% and accelerate model convergence during training.
- Conducted rigorous model evaluation, hyperparameter optimization, and ablation studies, establishing highly reliable baselines and improving overall inference accuracy by 18% for downstream analytical tasks.
- Took the product from zero to a working MVP in 3 months, owning the organizer dashboard, data model, and deployment pipeline.
- Designed a Trip Template + Cohorts data model enabling organizers to manage dozens of departures, prices, and occupancies per template with zero data duplication.
- Built a custom abstraction layer over Firebase/Firestore providing ACID-like transactional guarantees and schema validation on a NoSQL store (no ORM available).
- Shipped lead capture, booking, and payment flows that replaced manual spreadsheet workflows and readied the product for Razorpay and WhatsApp integrations.
- Built a reusable Next.js + Tailwind component library, cutting new dashboard view delivery from days to hours.
Skills
Languages
- Python
- TypeScript
- JavaScript
- SQL
- C++
- Java
AI / ML / DL
- PyTorch
- TensorFlow
- Transformers
- LangChain
- OpenAI API
- Scikit-learn
Backend
- Node.js
- FastAPI
- Express
- REST APIs
Frontend
- React
- Next.js
- Tailwind CSS
Database & Cloud
- PostgreSQL
- MongoDB
- Redis
- AWS
- Vercel
MLOps & Deployment
- Docker
- CI/CD for ML
- Model Serving (vLLM, TorchServe)
- MLflow
- Weights & Biases
Achievements
92.1 percentile
All India rank 5,472 of 69,242
Data Science and Artificial Intelligence (DA)
GitHub Activity
Let's Connect
Have a project in mind, or just want to talk? My inbox is open and I answer everything.
- MAILanikajain1307@gmail.com
- GITHUBgithub.com/anidoesdev
- LINKEDINlinkedin.com/in/anidoesdev
- TWITTERtwitter.com/anidoesdev
© 2026 Anika Jain - designed and built from scratch
end of transmission