Work Projects Experience Skills Trainings Credentials Contact

Kathmandu, Nepal · open to research, consulting & freelance work

Data-driven FinTech researcher, innovator and trainer

I design and build innovations across FinTech — research systems, automated financial models and AI tools that turn slow, manual work into repeatable pipelines a team can actually audit. Five of my AI projects are live and free to try, including price estimators trained on thousands of real Nepali listings. Alongside consulting work, I train finance professionals to use AI and automate their own workflows.

View FinTech work Try live projects
Biggyat Kumar Pandey

Confidential FinTech work

Laxmi Sunrise Capital

DPR Risk Reporting & Short-Term Stock Assessment

Daily risk report, automated end to end

Automated the daily risk report that used to be assembled by hand, so the desk gets volatility numbers before the market opens.

Full case study
Problem
Manual tracking of short-term volatility and portfolio risk exposure was too slow for dynamic market conditions.
Objective
Develop a daily automated reporting framework to quantify short-term stock risks.
Methods
Time-series analysis, statistical volatility modeling, and automated data aggregation.
Impact
An automated reporting engine that significantly accelerated daily investment decision support.

Python · Financial Engineering · Risk Analysis

Laxmi Sunrise Capital

Financial Sector Grading System

9 economic sectors on one scale

One scoring model that ranks companies consistently across nine sectors, so a bank and a hydropower firm can actually be compared.

Full case study
Problem
No standardized, quantitative way to evaluate and compare companies across very different economic sectors.
Objective
A unified grading system covering nine economic sectors, based on fundamental metrics.
Methods
Weighted scoring algorithms, fundamental ratio normalization, and cross-sectional analysis.
Impact
Gave portfolio managers a quantitative ranking framework, streamlining capital allocation.

Fundamental Analysis · Algorithm Design · Sector Research

Laxmi Sunrise Capital

LS Fund Dashboard

2 teams working from one app

Replaced scattered fund spreadsheets with one Streamlit app the research and portfolio teams open every day.

Full case study
Problem
Research insights and fund metrics were siloed in static spreadsheets, making real-time portfolio monitoring difficult.
Objective
Centralize portfolio analytics and research data in a single, interactive application.
Methods
Data visualization, interactive state management, and backend database integration.
Impact
A Streamlit dashboard used directly by the research and portfolio management teams.

Streamlit · Python · Data Visualization

More projects

Research Pipelines for Financial Data The ETL layer underneath everything else — cleans, validates and harmonises messy NEPSE listed-company data.
Problem
Inconsistent data quality and missing values in raw financial datasets skewed analytical models.
Objective
A robust ETL architecture for cleaning and harmonizing Nepalese financial data.
Methods
Data imputation, anomaly detection, schema validation, and SQL database management.
Impact
A clean, reliable data warehouse that serves as the foundation for all the quantitative models.

SQL · ETL · Data Engineering

Capital Market Research Automation Tracks macro indicators on a schedule and writes the daily and weekly market briefing itself.
Problem
Macroeconomic indicators and market-wide trends were not systematically tracked in an easily accessible format.
Objective
Python infrastructure to systematically monitor and analyze broad capital market trends.
Methods
API integration, statistical trend analysis, and automated report generation.
Impact
A suite of automated scripts that generate daily and weekly market intelligence briefings.

API Integration · Market Intelligence · Python

Equity & Fundamental Research Automation Pulls the numbers out of quarterly PDFs automatically, turning days of manual extraction into a script run.
Problem
Fundamental research at ShareSanskar required hundreds of hours of manual data extraction from quarterly reports.
Objective
Automate the extraction and calculation of fundamental metrics for rapid market reporting.
Methods
Data wrangling, automated web scraping, and programmatic PDF parsing.
Impact
End-to-end Python scripts that drastically reduced manual effort and accelerated reporting.

Python · Data Extraction · Automation

Live projects

Open-source work you can use right now, free and with no sign-up. The AI models run in your browser, and each repository documents its data, method and test results.

Built for my current organisation · TaskNTrade

TaskNTrade: creator-led marketing platform

Live company website · brands & creators

The website for Task n Trade in Lalitpur, which matches brands with UGC creators on delivered performance. It has separate journeys for brands and creators, a campaign explorer, services, market insights and a two-minute creator application.

Visit taskntrade.com

Real estate · Nepal

Nepal Property Price Estimator

1,900+ real listings · 111 localities

Estimates the asking price of a house or land anywhere from Kathmandu to the Terai. Works in ropani-aana or bigha-kattha-dhur, shows a likely range and price per aana, and lists comparable properties on the market.

Web scraping · Hedonic regression · JavaScript

Live demo Source

Automobiles · Nepal

Nepal Used Car Value Estimator

68 models · ±7.7% typical error

Values a used car from real Hamrobazar listings, compares it with today's showroom price, and shows how fast each model loses value. Handles Bikram Sambat model years.

Data collection · Ridge regression · Depreciation modelling

Live demo Source

Medical AI · Research

MediQNet: Medical Visual Q&A

60.2% on VQA-Med 2019 · would rank #5 of 18

Ask a question about a radiology image (modality, plane, organ system or abnormality) and the network answers on your device. Plane accuracy beats the 2019 challenge winner.

PyTorch · Computer vision · ONNX in the browser

Live demo Source

Sports analytics

Premier League 2020/21 in Numbers

488 players · 13 leaderboards

An interactive season dashboard: leaderboards, a shots-versus-goals finishing chart, which team stats actually track with winning, and percentile profiles to compare any two players.

Data visualisation · R · JavaScript

Live demo Source

Education analytics

Student Performance Insights

R² 0.88 on unseen students

Predicts a student's maths score from background and reading and writing results, explains which inputs moved each estimate, and shows what 1,000 students' results reveal.

ML pipeline · Explainable AI · CI/CD

Live demo Source

E-commerce · VIT team project

Farmkart: Online Store for Farmers

117 products · seasonal sowing guide

An online store for seeds, plants, tools and pest control, with guidance on what to sow each season. Originally a PHP and MySQL web app; the live version lets you search, filter and fill a cart.

PHP · MySQL · JavaScript

Live demo Source

Everything in one place

Project hub

Every live demo, repository and write-up on one page, with academic reports and earlier experiments included.

Open the hub GitHub profile

Experience

  1. Jul 2026 — Present

    TaskNTrade Private Limited

    Head of Research and Development

    Leading research and development in analytics and secure, intelligent systems built with AI. Developed the company website, taskntrade.com, for its creator-led UGC marketing platform.

  2. Jun 2026

    ShareSanskar

    Researcher

    Equity and fundamental research for one of the leading finnews portals in the market — including Budget 2084 analysis and data-driven coverage of listed companies and the macroeconomic indicators moving them.

  3. 2024 — 2026

    Laxmi Sunrise Capital Limited

    Junior Research Analyst

    Ran quantitative equity research and fundamental analysis for institutional portfolio management, and rebuilt the parts of that workflow that were still manual as Python pipelines the research team now runs daily — including the three frameworks under Selected work.

    The attestation covers my engineering and research contributions only. All intellectual property remains with Laxmi Sunrise Capital, and no proprietary code, data or model output is reproduced on this site.

  4. Apr — Jun 2024

    SmartInternz

    Cloud & Machine Learning Externship

    Built and shipped a car-performance prediction model end to end: a Flask service behind Gunicorn, containerised with Docker and deployed to AWS through a GitHub Actions CI/CD pipeline. I later rebuilt the project for the Nepali market as the used-car value estimator above.

    Try the car value estimator
  5. 2020 — 2024

    Vellore Institute of Technology

    B.Tech Computer Science — AI & Data Science

    MediQNet, my featured research: a deep-learning system that answers radiology questions straight from the image, trained on the ImageCLEF VQA-Med-2019 benchmark (3,200 images, 12,792 question–answer pairs). A compact version now runs in the browser and scores 60.2% on the official test set, a score that would have placed 5th of 18 teams in the 2019 challenge.

    Try the live demo Read the project reports

    Other projects: IoT smart energy management, a network analysis of racial bias in football magazine coverage, a colour-recognition tool for hands-free cursor control, and PYCHAIN, a minable cryptocurrency built from scratch.

Core competencies

What I can take on, grouped by the areas I've actually worked in — from equity research desks and creator marketing to energy modelling and the training room.

Capital markets & equity research

Quantitative and fundamental research for institutional portfolio management and public market coverage — scoring, risk and macro analysis a desk can act on.

Used at Laxmi Sunrise Capital · ShareSanskar

  • Equity research
  • Fundamental analysis
  • Financial modeling
  • Sector research & grading
  • Portfolio risk reporting
  • Volatility analysis
  • Macroeconomic & budget analysis
  • NEPSE market data

Creator marketing & brand intelligence

Architecting AI solutions for creator-led marketing: matching brands with creators on delivered performance rather than follower count, and turning every campaign into intelligence a brand can act on — who to brief, what to fund and what to scale.

Used at TaskNTrade · creator-led UGC campaigns · taskntrade.com

  • Creator intelligence & scoring
  • Brand intelligence
  • Creator–brand matching
  • Creator marketing solutions
  • Campaign performance analytics
  • CPV & engagement benchmarking
  • Audience & sector insights
  • Content performance analysis
  • Marketing data pipelines
  • AI solutions architecture

Data engineering & automation

Turning manual, spreadsheet-bound work into pipelines and dashboards that run on a schedule and can be audited line by line.

Used at Laxmi Sunrise Capital · ShareSanskar · TaskNTrade · Nepal price estimators

  • Python
  • SQL
  • ETL pipelines
  • Data validation & cleaning
  • Web scraping (robots.txt-compliant)
  • PDF data extraction
  • API integration
  • Automated reporting
  • Streamlit dashboards
  • Data visualization

AI & machine learning

From research models to AI-assisted workflows in production — including computer vision for medical imaging and emerging tech built from first principles.

Used in MediQNet · Nepal property & car estimators · AI-assisted research at Laxmi Sunrise · SmartInternz · VIT projects

  • Deep learning
  • Computer vision
  • Visual question answering
  • Predictive modeling
  • Pricing & valuation models
  • Explainable AI
  • AI-assisted research workflows
  • AI tools for finance
  • Social network analysis
  • Blockchain fundamentals

Cloud & deployment

Shipping models as real products: containerised services, or models exported to run in the browser so anyone can use them free, with no server to maintain.

Used at SmartInternz · five live projects on GitHub Pages

  • Flask
  • Gunicorn
  • Docker
  • AWS
  • GitHub Actions CI/CD
  • GitHub Pages
  • ONNX & in-browser inference
  • Model serving

Energy & infrastructure

Modelling for hydropower and solar, infrastructure risk evaluation, and real-time energy monitoring from sensor data.

Used in ChesCo energy table · IoT smart energy project at VIT

  • Energy modeling
  • Hydropower & solar analysis
  • Infrastructure risk evaluation
  • IoT sensor data
  • Energy monitoring & optimisation

Training & facilitation

Workshops that teach the algorithm and the business case together, built around live prototypes the room can inspect.

Used at ACCA Nepal · 200+ professionals trained

  • Workshop facilitation
  • AI for finance professionals
  • Python for finance
  • Curriculum design
  • Live prototyping
  • Public speaking

Trainer & workshop facilitator experience

ACCA Nepal speaker sessions · May — July 2026

Taught corporate finance professionals how to fit AI pipelines onto legacy accounting systems without rebuilding them. Sessions run on live Python prototypes — automated financial models the room can inspect line by line — and cover where automation earns its keep and where a human still has to sign off.

200+ Professionals trained
5 Core tech stacks taught
See the sessions on LinkedIn

“A model is only worth what the person reading it can act on.”

Credentials

Letters of recommendation, attestations and certificates. Click a document to read it full size.

Let's talk

If you're trying to automate a research workflow, grade a sector, or get a team off spreadsheets — I'd like to hear about it. WhatsApp is fastest; I usually reply the same day.

Message on WhatsApp Email LinkedIn GitHub All projects