∫
loading experience
ekansh dave / experience

experience

Research, independent study, and competitive achievements — a record of where I've worked, what I've built, and what I've earned.

// TIMELINE

research & work

2024 — PRESENT
JOHNS HOPKINS UNIVERSITY
Research Assistant — Materials Science & Engineering
Department of Materials Science and Engineering · Baltimore, MD
Undergraduate researcher in the Hopkins MSE department. Focus on computational simulation of crystalline defect dynamics — using mathematics and computation to understand how materials fail at the atomic scale.
  • Developed Python simulation pipelines to model dislocation formation and propagation in crystalline lattices under applied stress fields
  • Applied linear algebra (tensor decomposition, eigenvalue analysis of stress tensors) and differential equations to represent grain boundary migration dynamics
  • Built data analysis tooling to process molecular dynamics simulation outputs — identifying defect nucleation sites from large coordinate datasets
  • Contributed to research connecting microstructural features to bulk mechanical properties, relevant to materials failure prediction
  • Presented findings in group lab meetings, developing scientific communication skills in a rigorous research environment
materials science python computational modeling linear algebra molecular dynamics data analysis differential equations
2023 — PRESENT
INDEPENDENT RESEARCH
ML Architectures from First Principles
Self-directed
Self-directed deep dive into machine learning theory — not library-level but mathematical-foundation-level. Goal: understand every parameter update, every loss surface, every attention weight from the math up.
  • Implemented a full neural network library in NumPy — forward pass, backpropagation via chain rule, SGD/Adam/RMSProp, batch normalization, dropout, from scratch
  • Implemented the Transformer architecture (Vaswani et al. 2017) from the paper's mathematics — multi-head attention, positional encoding, layer normalization, FFN layers
  • Derived and verified the connection between Lagrangian dynamics and neural network optimization via Hamiltonian mechanics analogies
  • Produced mathematical writeups connecting implementation to theory — available as project documentation
deep learning backpropagation transformers numpy attention mechanisms
2023 — PRESENT
NATIONAL COMPETITIVE DEBATE
Policy & LD Debater
NFHS / Varsity Circuit
National-level policy and Lincoln-Douglas debater. Debate is applied rigorous argumentation — the same intellectual discipline as proof construction, applied under time pressure to complex policy questions.
  • State-qualified in NFHS Policy — 3-2 prelims, octofinals at State Qualifier
  • Finalist at Winter Classic Invitational PF (5-0 prelims, AI regulation topic)
  • Semifinalist at Regional LD Invitational, predictive policing topic
  • Research methodology: academic journals, RAND/Brookings/CSIS, evidence cutting with argument structure tagging
policy debate lincoln-douglas public forum research
2022 — PRESENT
COMPETITION MATHEMATICS
AMC / AIME Competitor
MAA AMC Program
Systematic preparation and competition in the AMC 10/12 and AIME pipeline. Focus areas: combinatorics, number theory, geometry, and proof-based problem solving.
  • AIME qualifier through AMC 12A and AMC 12B — top 5% nationally
  • Self-study curriculum: AoPS Volume 1 & 2, USAMO preparation, proof-writing practice
  • Strong emphasis on combinatorics and number theory — areas with direct application to CS and cryptography
AMC 12 AIME number theory combinatorics
// ACHIEVEMENTS

selected record

2025 · MATHEMATICS
AIME Qualifier
AMC 12A / 12B · MAA
Qualified for the American Invitational Mathematics Examination through both AMC 12 sittings. Top 5% nationally.
NATIONAL
2025 · DEBATE
State Qualifier
NFHS Policy Debate
Qualified for state championship through district rounds. 3-2 prelims, octofinal elimination at state qualifier.
STATE
2025 · DEBATE
Tournament Finalist
Winter Classic PF Invitational
Undefeated prelims (5-0) on the AI regulation topic. Lost finals in a close split decision.
FINALIST
2025 · RESEARCH
JHU Research Appointment
Materials Science · Johns Hopkins
Selected for undergraduate research position in Hopkins MSE department. Computational defect modeling.
RESEARCH
2024 · DEBATE
LD Semifinalist
Regional Invitational · LD
4-1 prelim record on the predictive policing topic, reached semifinals of regional invitational.
REGIONAL
2023–24 · ML
Neural Net from Scratch
Independent Project
Complete deep learning library in pure NumPy — backprop, Adam, batch norm, all from mathematical derivation.
TECHNICAL
// COURSEWORK & STUDY

relevant coursework

MATHEMATICS
Multivariable Calculus
Partial derivatives, vector fields, Stokes
MATHEMATICS
Linear Algebra
Eigensystems, SVD, matrix decompositions
MATHEMATICS
Differential Equations
ODEs, PDEs, stability analysis
MATHEMATICS
Probability & Statistics
Measure-theoretic foundations
COMPUTER SCIENCE
Data Structures & Algorithms
Complexity, graphs, dynamic programming
COMPUTER SCIENCE
Cryptography
RSA, ECC, hardness assumptions, protocols
COMPUTER SCIENCE
Systems Programming
C, memory, OS fundamentals
PHYSICS
Classical Mechanics
Lagrangian & Hamiltonian formalism
PHYSICS
E&M / Waves
Maxwell's equations, waveguides
SELF-STUDY
Deep Learning Theory
Goodfellow, Sutton & Barto, Vaswani
SELF-STUDY
Competition Math (AoPS)
AMC/AIME/USAMO preparation
SELF-STUDY
Materials Science
Callister, crystallography, defect theory