Akshay Mallireddy
Available for internships & research

Hi, I'm Ramaakshay

Aerospace Engineering @ UT Austin

Building at the intersection of aerospace, software, and machine learning — from orbital simulations and spacecraft CAD to robotic control systems and cloud infrastructure.

Coding Projects

IBM WorkTrace

Overview

Built with a founder mindset at IBM — product managed the full lifecycle from market analysis of intern pain points through production deployment. A time reporting and billability tracking platform adopted by interns and managers across IBM.

Technical Details

  • RAG platform with semantic search, NLP-powered meeting summarization, and multi-provider LLM inference (watsonx.ai, Groq, Ollama)
  • Agentic pipeline that parses transcripts, classifies tasks, and routes time entries into PostgreSQL autonomously
  • Integrated with agentic coding tools for in-editor work logging
  • Deployed on OpenShift with Supabase-managed Postgres, environment-driven provider switching, and GitHub CI/CD

Stack & Skills

  • Python
  • FastAPI
  • React
  • pgvector
  • LangChain
  • watsonx.ai
  • OpenShift
  • PostgreSQL
WorkTrace

Real-Time Stock Tracker

Overview

A real-time stock dashboard built in Next.js/TypeScript featuring live TradingView charts, market heatmaps, sector-grouped quotes, and financial news — iterated on based on user feedback and customer empathy.

Technical Details

  • TradingView widget integration for live charts, market heatmaps, and sector-grouped quotes
  • Personalized watchlists, configurable price alerts, and per-ticker financials
  • Modular React/TypeScript component architecture for maintainability and extensibility
  • Iterated features based on direct user feedback and continuous usability testing

Stack & Skills

  • Next.js
  • TypeScript
  • React
  • Tailwind CSS
  • TradingView API
Stock Tracker

Hybrid Anomaly Detection — IBM

Overview

Engineered a hybrid STL + Isolation Forest model for time-series pipeline metrics at IBM, using statistical seasonal-trend decomposition to isolate residuals before ML-based scoring to cleanly distinguish true anomalies from normal seasonality.

Technical Details

  • STL decomposition separates seasonal, trend, and residual components from raw time-series data
  • Isolation Forest applied to residuals for unsupervised anomaly scoring
  • Prevents false positives caused by recurring seasonal patterns in pipeline metrics
  • Deployed as part of cloud-native monitoring infrastructure on Red Hat OpenShift

Stack & Skills

  • Python
  • scikit-learn
  • Isolation Forest
  • STL Decomposition
  • Time-Series Analysis
  • OpenShift
Anomaly Detection

IL & RL Blended Recovery Architecture

Overview

A dual-policy robotic control framework combining imitation learning (behavior cloning) with deep reinforcement learning to handle out-of-distribution states in manipulation tasks. Built as part of the Robot Learning FRI research program at UT Austin.

Technical Details

  • Dual-policy architecture: behavior cloning IL policy for nominal operation + actor–critic RL recovery for OOD states
  • Safety-constrained action projection (shielding layer) bounds control inputs and prevents collisions
  • Leveraged NLP to parse and summarize experiment results across training runs
  • Evaluated on simulated manipulation tasks — tracked collision rate, recovery latency, and task success

Stack & Skills

  • Python
  • PyTorch
  • Reinforcement Learning
  • Behavior Cloning
  • NLP
  • Robotics Simulation
Robot Learning

Drone Network — Swarm RL

Overview

A multi-agent reinforcement learning system that trains a swarm of mini-drones to cooperatively complete household tasks — watering plants, sweeping floors, and toggling lights. Trains on a fast pure-Python simulation and deploys into a PyBullet physics lab with real Crazyflie quadrotor aerodynamics.

Technical Details

  • MAPPO (Multi-Agent PPO) with Centralised Training, Decentralised Execution (CTDE) — shared actor per drone + global central critic
  • Pure-Python/NumPy training environment with a parallel PyBullet physics deployment env for real quadrotor dynamics
  • GAE rollout buffer, PPO clip update, and running reward normalisation with curriculum scheduler
  • Cross-platform installer that auto-patches PyBullet source on macOS Sequoia/Tahoe (clang 17+ / SDK 15+)

Stack & Skills

  • Python
  • PyTorch
  • Multi-Agent RL
  • MAPPO
  • PyBullet
  • NumPy
Drone Network

Orbital Energy Simulation — Texas Spacecraft Lab

Overview

Developed orbital energy simulations for spacecraft power system validation as part of the Texas Spacecraft Laboratory's Electrical Power Systems sub-team. Designed EGSE PCBs and validated power models against real load profiles and bench data.

Technical Details

  • Designed EGSE PCB integrating complex components for spacecraft power system testing and validation
  • Selected and tested 15+ power components — batteries, solar panels, and converters
  • Ran orbital simulations modeling power consumption across Earth-pointing and maneuvering mission profiles
  • Validated simulation results against load profiles and bench measurement data

Stack & Skills

  • MATLAB
  • Orbital Mechanics
  • PCB Design
  • Power Systems
  • Spacecraft Engineering
Spacecraft Lab

CAD Projects

Autonomous Navy Fighter Jet — AIAA

Objectives

Design a next-generation autonomous fighter jet for the Navy that breaks Augustine's Law while achieving Mach 1.8 speeds and incorporating stealth capabilities. Submitted for the AIAA Aircraft Design Competition.

Technical Details

  • Revolutionary cost-reduction approach while maintaining Mach 1.8 performance
  • Advanced autonomous flight control systems for fully unmanned combat operations
  • Stealth integration: radar-absorbent materials and reduced-signature geometry
  • Modular architecture for cost-effective manufacturing, maintenance, and carrier compatibility

Skills & Tools

  • Aerospace Systems Eng.
  • Autonomous Systems
  • Supersonic Aerodynamics
  • Stealth Technology
  • Cost Analysis
  • CAD
Autonomous Navy Fighter Jet

F-18 Remastered

Objectives

Redesign and refine a high-fidelity CAD model of the F-18 to improve geometric accuracy, surface continuity, and subsystem realism while maintaining structurally plausible aircraft features.

Technical Details

  • Rebuilt geometry using parametric surface and solid modeling for accurate fuselage curvature and wing–body blending
  • Modeled wings, stabilators, vertical tails, and engine inlets with realistic proportions
  • Feature-based design for control surfaces enables iterative refinement
  • Clean topology and surface continuity suitable for rendering or future CFD meshing

Skills & Tools

  • SolidWorks
  • Fusion 360
  • Surface Modeling
  • Aircraft Geometry
  • Aerospace Design
F-18 Remastered

CubeSat with Liquid Propulsion

Objectives

Design a CubeSat platform integrating a compact liquid propulsion system for controlled orbital maneuvering and attitude adjustments within CubeSat mass, volume, and power constraints.

Technical Details

  • Subsystem-level CAD architecture integrating propulsion, structure, avionics, and payload within CubeSat form-factor
  • Liquid propulsion layout with tanks, feed lines, valves, and thruster placement for CoM stability
  • Internal volumes allocated for power, communication, and control subsystems
  • Modular architecture supporting Earth-pointing, maneuvering, and attitude control mission modes

Skills & Tools

  • Fusion 360
  • SolidWorks
  • CubeSat Systems Eng.
  • Liquid Propulsion
  • Spacecraft Design
CubeSat CAD

Headphone Stand

Objectives

Design a compact, ergonomic headphone stand with an integrated AirPods Pro holder optimized for desktop use and 3D printing.

Technical Details

  • Modeled in Fusion 360 with parametric dimensions for print tolerance
  • Optimized overhangs and geometry for FDM printing without support structures
  • Integrated dedicated AirPods Pro dock into the base design

Skills & Tools

  • Fusion 360
  • Parametric Modeling
  • Design for Manufacturing
  • FDM 3D Printing
Headphone Stand CAD

AIAA Competition Report

Full design report submitted for the AIAA Aircraft Design Competition — University of Texas at Austin, 2026.

View Full Report

Coursework

Completed

  • Multivariable CalculusM408D
  • AI Design and DevelopmentITD111
  • Computer Programming (Python)CS303E
  • C++ Data Structures & AlgorithmsCOE301
  • Differential Equations & Linear AlgebraM427J
  • StaticsE M 303
  • ThermodynamicsME 310T
  • Robot LearningCS 309
  • Engineering ComputationCOE 311K

Current

  • Vector CalculusM427L
  • DynamicsE M 311M
  • Mechanics of SolidsE M 319
  • Robot Learning FRI IIC S 378
  • Engineering CommunicationE S 333T

Resume

Aerospace Engineering student with experience in ML, embedded systems, cloud infrastructure, and spacecraft design.

Download Resume

Contact

Let's connect

Open to internships, research collaborations, and interesting engineering problems. Reach out any time.

akshaymall@utexas.edu
Akshay's AI Twin
Online
Hi! I'm Akshay's AI twin. Ask me anything about his background, skills, or projects.