Looking for a job

Jinia Haldar

@jiniahaldar

Message

I build energy forecasting models, QA automation, and reliable AI training workflows.

United States

Message

What I'm looking for

I am looking for a quantitative role focused on building rigorous, trustworthy models for complex systems. My priority is joining a team that values first-principles statistical thinking over black-box tool usage, gives engineers ownership over data validation and pipeline integrity, and uses data to drive real operational decisions rather than superficial metrics.

I've improved energy-market forecasting products at Aurora Energy Research, serving as the primary modeling SME for the Alberta product. I improved Alberta price delta by 50% and runtime by 20% using NumPy, Pandas, and Git-managed development.

I reduced manual QA overhead by 50% across 20 forecasting products by building Python automation tools and Power BI dashboards using DAX and SQL. I also modeled CCGT dispatch for NORAM and contributed modeling expertise to three Advisory projects.

As an AI Data Trainer, I perform RLHF for STEM-focused LLMs, checking physics, calculus, and linear algebra reasoning. I identify and correct hallucinations in multi-step reasoning chains to improve model truthfulness.

Earlier, I built a TensorFlow CNN for microscopy autofocus at Malvern Panalytical, achieving an R² of 0.96. I hold a first-class MSci in Physics and completed a thesis on automated Python image analysis of semiconductor wafer images.

Experience

Work history, roles, and key accomplishments

Current

AI Data Trainer

Current

DataAnnotation.tech

Jan 2025 - Present (1 year 8 months)

Performed RLHF for STEM-focused LLMs, verifying model accuracy in physics, calculus, and linear algebra. Identified and corrected hallucinations in multi-step reasoning chains to improve model truthfulness.

RLHF LLM Evaluation Physics Calculus Linear Algebra Reasoning Data Annotation Reinforcement Learning from Human Feedback (RLHF)Machine Learning Quality Assurance

Energy Modeling Analyst

Aurora Energy Research

Sep 2022 - Aug 2024 (1 year 11 months)

Primary modeling SME for Alberta product; improved Alberta price delta by 50% and runtime by 20% by implementing features using NumPy and Pandas within a Git-managed workflow. Reduced manual QA overhead by 50% for 20 forecasting products by developing Python-based automation tools and Power BI (DAX/SQL) dashboards. CCGT dispatch modeler for the NORAM team, providing modeling expertise for three Advisory projects.

Python NumPy Pandas Git Power BI DAX SQL Energy modeling Forecasting CCGT Data Analysis Automation Energy Forecasting CCGT Dispatch Modeling Quality Assurance Automation

Placement Year / Internship

Malvern Panalytical

Sep 2020 - Sep 2021 (1 year)

Independently built a convolutional neural network (CNN) using TensorFlow for microscopy autofocus. Achieved an R² of 0.96 between predicted and expected values, exceeding the initial target of 0.85.

TensorFlow CNN Microscopy Autofocus Python Machine Learning Convolutional Neural Networks (CNNs)

Education

Degrees, certifications, and relevant coursework

University of Bristol

Integrated Master of Science (MSci), Physics

2018 - 2022

Grade: First-class honors (4.0 GPA)

Awarded first-class honors (4.0 GPA) in a 4-year combined undergraduate and graduate program. Thesis: Automated image analysis in Python of semiconductor wafer images taken with EUV lithography.