Praneeth Tota

Staff Software Engineer · AI Systems · Distributed Infrastructure
Sunnyvale, CA (773) 620-0036 praneeth200@gmail.com LinkedIn GitHub praneethtota.github.io

Staff Software Engineer with 10+ years of industry experience and 6 years of doctoral research, designing and operating large-scale distributed systems across pricing, supply chain, and data platforms at Apple and Walmart. Proven track record of owning critical system components end-to-end, driving cross-team technical decisions, and delivering high-reliability platforms under strict performance and correctness constraints.

PhD in Computer Science (Algorithms & Game Theory) with a rigorous mathematical foundation in constrained optimization, primal-dual methods, Lagrangian relaxation, Nash equilibrium theory, and Stackelberg game analysis — the mathematical underpinnings of modern ML training, RLHF, and adversarial optimization. Thesis formally quantified how individual reward maximization degrades collective welfare in multi-agent systems, a problem structurally identical to reward misalignment in RL. Hands-on experience with PyTorch.

Global pricing platform — Apple worldwide Last-mile delivery — 4,000+ Walmart stores · 2-hr SLA Adaptive Utility Agents — open-source LLM control plane Streaming OLAP · Kafka → Flink → ClickHouse
Skills
Languages
Python, Java, SQL
Distributed Systems
Microservices, Event-driven architecture, REST, gRPC
Data & Streaming
Kafka, Flink, Spark, Redis, ClickHouse, ETL Pipelines
Cloud & Infrastructure
AWS, Kubernetes, Docker, Terraform
Databases
Oracle, DynamoDB, MongoDB, PostgreSQL, Cassandra, Elasticsearch
Observability
Splunk, ELK Stack, Prometheus, Grafana
ML & Mathematics
PyTorch, constrained optimization, primal-dual methods, linear algebra, Lagrangian methods, game theory
Experience
Senior Software Engineer / Tech Lead Contract
Apple — deployed via Infosys
Dec 2022 – Present
Sunnyvale, CA · Contractor (Infosys) embedded in Apple pricing engineering
Global pricing platform · Apple products worldwide
  • Owned design and evolution of the global pricing platform responsible for Apple product pricing across all regions, supporting high-stakes launch events with strict correctness and latency requirements.
  • Led migration of pricing systems from on-premises infrastructure to AWS, reducing annual downtime from ~48 hours to ~4 hours and significantly improving horizontal scalability.
  • Built automation and validation frameworks that reduced production incidents by over 90% and improved deployment reliability across all pricing services.
  • Designed and implemented Python + SQL ETL pipelines for data validation and reconciliation across upstream and downstream systems, reducing user-facing support tickets by 30%.
  • Architected a streaming OLAP pipeline (Kafka → Flink → ClickHouse) enabling real-time data processing and analytics across pricing workflows, replacing batch-based approaches.
  • Drove cross-team technical decisions on system design, release strategy, and reliability standards; collaborated with Finance, Data Science, and Platform teams to productionize pricing models.
Senior Software Engineer / Tech Lead Contract
Walmart — deployed via Infosys
Sep 2019 – Nov 2022
Sunnyvale, CA · Contractor (Infosys) embedded in Walmart delivery engineering
Last-mile delivery platform · 4,000+ stores · 2-hour SLA
  • Led reliability and performance improvements for a last-mile delivery platform operating under strict 2-hour SLA constraints across 4,000+ Walmart store locations.
  • Designed observability-driven debugging workflows using Splunk, ELK, Prometheus, and Grafana, cutting mean incident response time from 45 to 20 minutes and improving on-time delivery by 15%.
  • Diagnosed and resolved distributed system failures by analyzing service-level logs, request flows, and Cassandra data inconsistencies — implementing root-cause fixes rather than workarounds.
  • Managed and mentored a team of 4 engineers, maintaining 24/7 platform reliability and enforcing operational best practices.
  • Collaborated cross-functionally with Product, Data, and Finance teams to align technical roadmap with delivery KPIs and financial targets.
  • Contributed to the evolution toward an event-driven, microservices-based architecture, improving fault isolation and system scalability.
Ph.D. Researcher
Illinois Institute of Technology
2013 – 2019
Chicago, IL
  • Formally proved bounds on the Price of Anarchy in multi-agent resource allocation games — quantifying how selfish individual optimization degrades collective welfare, a problem structurally identical to reward hacking and misalignment in reinforcement learning systems.
  • Analyzed Stackelberg (leader-follower) games with concave utility functions and convex cost functions — the same mathematical framework underlying adversarial training, GAN objectives, and RLHF policy/reward-model dynamics.
  • Studied proportional sharing mechanisms mathematically analogous to softmax attention, the core operation in transformer architectures.
  • Designed and proved two novel evaluation metrics (Wealth Impact Factor and Economic Competition Factor) to measure fairness and efficiency loss — demonstrating the ability to formally define and bound what a system actually optimizes for.
  • Implemented optimization algorithms and experimental models in Python and PyTorch; secured grant funding for research on jitter reduction in multipath TCP.
  • Taught and mentored 1,000+ graduate students in algorithms and distributed systems, bridging theoretical foundations with systems engineering.
Software Engineer
Epoch Solutions
2010 – 2012
Chicago, IL
  • Designed and developed Mobile Swan, a geo-social network app and backend services using MVC architecture.
  • Built logistics optimization algorithms, improving delivery routing efficiency.
Open Source Projects
Adaptive Utility Agents Open Source
Independent Project  ·  github.com/praneethtota/Adaptive-Utility-Agent
2025 – 2026
praneethtota.github.io/Adaptive-Utility-Agent
  • Designed and built a utility-governed correction framework for deployed LLMs — a control plane that detects errors, stores verified fixes in a persistent assertions store, and prevents recurrence across sessions without waiting for model retraining.
  • Validated through controlled simulation: 69.6% reduction in repeated errors over an uncalibrated baseline (500-task, 5-cycle study); +10.5% correctness gain from VCG-arbitrated specialist routing (p = 0.029); Pearson r = 0.461 between utility score and ground-truth correctness (p < 10−40).
  • Designed automated blue-green deployment triggered by utility deviation — field-calibrated thresholds derived from DPO penalty multipliers; surgical submodels require ≥246 canary interactions before any traffic shift, creative writing promotes in 2.
  • Published full whitepaper (formal proofs: Debreu representation theorem, VCG mechanism design, Lyapunov stability), builder's tutorial with runnable code, and seven domain deep-dives covering autonomous vehicles, energy grids, pricing systems, and AI infrastructure.
Education
Ph.D. in Computer Science — Algorithms & Game Theory
Illinois Institute of Technology, Chicago, IL
Thesis: Price of Anarchy bounds in multi-agent resource allocation games. Formal proofs connecting individual reward maximization to collective welfare degradation — the mathematical structure underlying reward misalignment in RL and adversarial optimization in modern ML systems.
2019
M.S. in Computer Science
Illinois Institute of Technology, Chicago, IL
2010