Welcome to Ruthvik Nath Bandari's portfolio

Skills

What I work with, grouped by area. Search a topic, or filter by category. Linked skills open the project or role where I used them; the rest are skills I list without pointing to a single piece of work.

221 skills in 7 categories

  • AI and machine learning

    Frameworks, model architectures and the training practice behind them.

    Frameworks and libraries

    What I build and evaluate models with.

    Model architectures

    Segmentation, language and sequence models.

    Training and evaluation

    Losses, schedules, calibration and data-quality checks.

    • Transfer learning
    • Fine-tuning
    • LoRA
    • Focal and Dice-CE losses
    • Cosine annealing
    • Exponential moving average (EMA)
    • Mixed precision (BF16/AMP)
    • Temperature scaling
    • Test-time augmentation
    • MC-dropout
    • Thompson-sampling bandits
    • Cleanlab
    • Cleanvision

    Clustering and dimensionality reduction

    Finding structure in unlabelled data.

  • Data engineering and analytics

    Storage, processing, visualization and the data sources behind the analysis.

    Databases and storage

    Where the data lives and how it is shaped.

    • PostgreSQL
    • Row-level security
    • Alembic
    • SQLAlchemy
    • SQLite
    • DuckDB
    • Redis
    • Parquet
    • JSONL
    • PyArrow

    Processing and analysis

    Dataframes, numerical work and analytics practice.

    Visualization and BI

    Charts, dashboards and reporting tools.

    Scraping and parsing

    Collecting and cleaning documents at scale.

    • Web scraping
    • BeautifulSoup
    • httpx
    • Requests
    • aiohttp
    • Selenium
    • Crawl4AI
    • Docling
    • Tenacity
    • MinerU

    Data sources and APIs

    Public records and datasets I have worked with.

    • PubMed API
    • arXiv API
    • OpenFDA
    • ClinicalTrials.gov
    • IPEDS
    • College Scorecard

  • MLOps, HPC and cloud

    Reproducible training on shared clusters, deployment, and the tooling that keeps it honest.

    Compute and HPC

    Cluster and GPU training.

    Experiment tracking

    Versioning data, runs and results.

    • MLOps
    • MLflow
    • Weights & Biases
    • DVC

    Deployment and hosting

    Containers, hosting, cloud services and monitoring.

    Dev tooling and quality

    Packaging, linting, CI and testing.

    • uv
    • Makefile
    • pre-commit
    • GitHub Actions
    • CI/CD
    • Ruff
    • mypy
    • oxlint
    • Git
    • VS Code
    • pytest
    • Hypothesis
    • Vitest
    • Playwright
    • axe-core
    • Testing Library
    • Bun test

  • LLMs and agents

    Retrieval-augmented generation, answer verification, hosted and local models, and tool-using agents.

    Retrieval and RAG

    Search, ranking and checking answers against sources.

    Agents and tooling

    Tool-calling agents and the tools I use to build them.

    • Tool-calling agents
    • JSON-Schema tools
    • Multi-Agent Systems
    • LangChain
    • Claude Code
    • n8n
    • Gemini CLI
    • Agent Skills
    • Prompt engineering
    • Context engineering

    Model hosting and APIs

    Self-hosted runtimes and hosted model APIs.

    NLP libraries

    Classical and modern text processing.

  • Computer vision and medical imaging

    Segmentation, image quality and evaluation that checks for leakage.

    Libraries and tasks

    Image handling, segmentation and recognition.

    Metrics and quality

    Overlap, boundary, calibration and fairness measures.

    • Dice and IoU
    • HD95
    • NSD
    • Calibration (ECE)
    • Skin-tone fairness analysis (ITA)
    • Wagner staging

  • Web and software engineering

    The languages, services and interfaces that make a model usable.

    Languages and runtimes

    What I write models, pipelines and interfaces in.

    Backend and APIs

    Services, validation and access control.

    Frontend

    Frameworks, state and component libraries.

    • React
    • Next.js
    • Vite
    • Tailwind CSS
    • TanStack Router
    • TanStack Query
    • Zustand
    • React Hook Form
    • shadcn/ui
    • Radix UI
    • MUI
    • Mantine

    UI and motion

    Animation and 3D in the browser.

  • Security and governance

    Vulnerability and attack-path risk modelling, alongside privacy, data-quality and responsible-AI habits for sensitive data.

    Security analysis

    Vulnerability scoring, attack paths and quantum exposure.

    Governance and privacy

    Habits for handling sensitive data.

    • Responsible AI
    • Data-quality auditing
    • Privacy and tenant isolation
    • Governance documentation

Logos are trademarks of their respective owners.