SYS.INITBUILD.2026

PrajwalSallapalli

Software Engineer×ML Research Scientist

I build systems that think and software that scales.

Software engineer and ML researcher building high-performance systems, distributed infrastructure and generative AI models — from low-latency C++ and network packet processing to transformers and diffusion architectures.

San Francisco, CAAVAILABLE FOR SOFTWARE / AI / SYSTEMS ROLES
LeetCode
SYSTEMS × INTELLIGENCE
NODE_NET
10K+
Orders / sec
15–20%
Network throughput improvement
30%
Software reliability improvement
10%
LLM benchmark accuracy improvement
15%
Training-time reduction
500+
AI product users
01ABOUT

Engineering from packets to parameters.

My work spans production networking software, C++ systems engineering, Linux packet processing, distributed infrastructure, and machine learning research — from LLM architectures and neural network optimization to diffusion models and scalable AI applications.

The unifying thread is building systems where performance, reliability and intelligence matter. Whether optimizing packet throughput at the kernel boundary or scaling transformer training pipelines, I operate at the intersection of systems engineering and artificial intelligence.

C++ SystemsNetworkingDistributed InfraLLMsDiffusion ModelsPyTorchPerformance
02EXPERIENCE

System history.

ROLE_ACTIVE

RooGen AI

ML Research Scientist

May 2026 — Present

  • ▸Developed and evaluated novel LLM architectures.
  • ▸Improved benchmark accuracy by approximately 10% through architecture modifications, training optimization and model scaling experiments.
  • ▸Designed neural network architectures with PyTorch.
  • ▸Reduced training time approximately 15% using optimized computation graphs, parallelism and efficient data pipelines.
  • ▸Researched diffusion-based generative models.
  • ▸Reduced GPU power consumption approximately 5% while preserving model quality and inference performance.
  • ▸Built training, evaluation and benchmarking frameworks.
  • ▸Worked on fine-tuning, transfer learning and parameter-efficient adaptation.
  • ▸Performed hyperparameter optimization and model profiling.
  • ▸Worked with research and engineering teams on production ML deployments.
PyTorchPythonLLMsTransformersDiffusionGPUModel OptimizationFine-Tuning
03PROJECTS

Built close to the metal. Designed for scale.

01SYSTEMS / C++ / TRADING INFRASTRUCTURE

Low Latency Order Book Engine

A high-performance multithreaded C++ matching engine implementing price-time priority and performance-conscious concurrent execution.

10K+ orders/secSub-millisecond executionMultithreaded
C++ConcurrencyMultithreadingLow-Latency SystemsData StructuresBenchmarking
REPOSITORYCASE STUDY
02DISTRIBUTED SYSTEMS

Distributed Data Processing Framework

A distributed processing framework supporting concurrent task execution, workload balancing, scheduling, monitoring and runtime analytics.

Distributed SystemsConcurrencySchedulingRuntime MetricsScalability
CASE STUDY
03AI / TRANSFORMERS

GPT Implementation & Instruction Fine-Tuning

Implemented a GPT-style transformer architecture in PyTorch and built training, evaluation and inference pipelines while studying model performance and scalability.

PythonPyTorchTransformersTokenizationLLMsFine-Tuning
REPOSITORYCASE STUDY
04FULL STACK / GENERATIVE AI

AI Resume Builder

Production AI resume platform with LLM-powered generation, real-time rendering and cloud deployment serving 500+ users.

500+ usersLLM-powered workflowsProduction cloud deployment
ReactNode.jsExpressMongoDBAWSLLMsREST APIs
REPOSITORYCASE STUDY
View All Projects →
04RESEARCH

Teaching machines efficiently.

EXPERIMENT.LOG — SUMMARIZED METRICS
EXPERIMENTSTATUSΔ
LLM ArchitectureCOMPLETE+10% benchmark
Graph OptimizeCOMPLETE-15% train time
Diffusion PowerCOMPLETE-5% GPU power

* Summarized research metrics — not live production monitoring

FOCUS AREAS
LLM Architectures
Transformer Models
Diffusion Models
Neural Network Optimization
Fine-Tuning
GPU Efficiency
Model Evaluation
Hyperparameter Optimization
05STACK

Layered engineering architecture.

AI / RESEARCH
PyTorch · LLMs · Transformers · Diffusion · Fine-Tuning · GenAI
DISTRIBUTED / INFRASTRUCTURE
AWS · Docker · Kubernetes · CI/CD
SYSTEMS
Linux · Networking · Concurrency · Multithreading · Performance
CORE
C++ · Python · C · SQL · JavaScript
languages
C++PythonCSQLJavaScript
machine Learning
PyTorchLLMsTransformer ArchitecturesDiffusion ModelsDeep LearningFine-TuningModel EvaluationGenerative AI
systems
LinuxUNIXMultithreadingConcurrencyPerformance Optimization
networking
TCP/IPUDPRouting ConceptsPacket ProcessingNetwork Debugging
distributed Systems
ScalabilityFault ToleranceDistributed Architectures
debugging
gdbRuntime AnalysisProfilingLog AnalysisBenchmarking
cloud
AWS EC2AWS S3DockerKubernetesCI/CD
ENGINEERING DNA

Focus Areas

PERFORMANCE
SYSTEMS
AI / ML
DISTRIBUTED
NETWORKING
PRODUCT

Relative focus areas — not literal skill percentages

CORE
PRAJWAL
SYSTEMS
PERFORMANCE
INTELLIGENCE
DISTRIBUTED
06EDUCATION
Expected Dec 2026

California State University, East Bay

Master of Science

Computer Science

Jun 2023

BMS College of Engineering

Bangalore

Bachelor of Engineering

Electronics & Communication

CAREER TIMELINE
2019
BMS College of Engineering
2023
HPE Intern → Software Developer
2025
CSU East Bay — MS Computer Science
2025
Zume / Product Building
2026
RooGen AI — ML Research Scientist
NOW
Building systems × intelligence
DEVELOPER NETWORK
GitHubLinkedInLeetCode

LET'S BUILD
SOMETHING
THAT SCALES.

Software EngineeringSystems EngineeringDistributed SystemsAI / ML Research
Start a Conversation →
GitHubLinkedInEmail