Projects
Batching Optimization for Vehicle Production Scheduling
Work
Batched vehicles with similar production dates and features to improve the stability of the production schedule.
- Developed efficient heuristics to optimize vehicle production scheduling, achieving near-optimal solutions with a 99% reduction in computation time compared to full solutions.
- Ideated, implemented, and evaluated a novel algorithm based on hierarchical clustering and optimal leaf ordering for sorting orders so that any subset of neighbors in the list are similar in terms of features and production date.
- Implemented optimizations that improved production delays by 7.8x, reduced date ranges within batches by 2.9x, and enhanced feature similarity across vehicles by 11%.
- Personally presented these results and the algorithm's benefits to top executives.
Iterative Dynamic Knowledge Graph with Source Traceability
PersonalResearch
A self-updating knowledge graph system that iteratively extracts entities, states, attributes, and relationships from text, while dynamically canonicalizing and refining graph structures as new data is provided, maintaining precise references to the source texts for accurate citation and verification.
- Designed a framework for accurate synthetic data generation adhering to specifications defined for each step in the pipeline.
- Engineered prompts to maximize performance on each step in the pipeline.
- Developed an evaluation framework to benchmark the performance of each step in the pipeline.
- Fine-tuned LLMs to combine steps in the pipeline resulting in an uplift of _ and a cost reduction of _
- Used reinforcement learning to train the answering engine LLM for knowledge graph traversal resulting in a Q&A improvement of _.
Artificial Intelligence Algorithm Implementations
Personal
Core AI algorithm implementations in Python with basic use cases.
I implemented the following concepts/algorithms and wrote blog posts about them:
- The Transformer Architecture for translation (encoder-decoder) and text-generation (decoder-only)
- Clustering - Partition-Based (K-Means), Hierarchical (Agglomerative and Divisive), Density-Based (DBSCAN), Model-Based (Gaussian Mixture Models)
- Linear Regression
- Logistic Regression
- Decision Trees: Basic ID3, CART, Random Forest, Gradient Boosted (XGBoost)
- Backpropogation
- Perceptron
- Gaussian Graphical Models
This project is a work in progress and 'm continuously adding new implementations as I learn about or review the respective algorithms.
Machine Learning Visualizations
Personal
Interactive visualizations for core machine learning concepts/algorithms.
I created visualizations for the following concepts/algorithms for some blog posts I wrote:
- Maximum Likelihood Estimation
- Bayesian Inference
- Principle Component Analysis
- Expectation Maximization
This project is a work in progress and I'm continuously adding new visualizations as I learn about or review ML algorithms/concepts.