w
whosreshuu

RISHANK SINGH

@whosreshuu

AI ML Engineer for Python Automation, Web Scraping and LLM Apps

India
Inglese, Hindi
Alcune informazioni sono riportate in lingua inglese.
Chi sono
AI/ML engineer who builds Python systems that run in production. I built a news-monitoring and LLM summarisation engine that a New York investment fund still uses weekly, and a satellite-imagery segmentation model (89% accuracy) for a Government of India solar project. I can help with: - Web scraping and automated data pipelines - LLM apps, summarisation and text classification - Computer vision and image segmentation - Data cleaning, analysis and automated PDF/Word/Excel reports Clear communication, clean code, on-time delivery.... Continua a leggere

Competenze

w
whosreshuu
RISHANK SINGH
offline • 
Tempo di risposta medio: 1 ora

Consulta i miei servizi

Sviluppo di chatbot IA
I will build python ai agents and telegram, discord or slack automation bots
A partire da40 USD / progettoMaggiori dettagli
API e integrazioni
I will build python rest apis with fastapi and integrate third party apis
A partire da35 USD / progettoMaggiori dettagli

Esperienza lavorativa

Bhaskaracharya_Institute For Space Applications and Geo-Informatics

AI/ML Intern

Bhaskaracharya Institute For Space Applications and Geo-Informatics

May 2026 - Jul 2026 • 2 mos

Owned the semantic-segmentation workstream of the GatiShakti PM-Surya Ghar Yojana under MeitY. Built a training corpus of 5,000 satellite image-mask pairs using OSMnx and Rasterio. Trained U-Net and DeepLabV3+ architectures in PyTorch, achieving 89% accuracy on live imagery. Developed an automated data correction step that reduced validation loss by 25%.

Plural_Investing LLC

AI Research Engineering Intern

Plural Investing LLC

Mar 2026 - May 2026 • 2 mos

Took an open-ended brief to a production system the fund still runs: an automated news-monitoring engine covering a 112-company research watchlist across US, UK, EU, Nordic, Israeli, Canadian and Australian listings. - Multi-source discovery and extraction from news RSS and regulated market feeds (SEC EDGAR, ASX, RNS, SEDAR+ and more) with headless Selenium, BeautifulSoup and trafilatura - Relevance layer: materiality classification, noise rejection and near-duplicate collapse across outlets - Benchmarked LLM summarisers (Llama 3.2 via Ollama, OpenAI, Anthropic, BART) behind a swappable provider interface with an offline fallback - Ships a weekly source-linked investor digest as PDF and DOCX (ReportLab, python-docx)