SINCE 2025
Perplexity
AI/ML Engineer
- Research-to-Production Infrastructure
- Foundation Models for Science
- Optimization Measured in Deltas
AI/ML Engineer · Perplexity · SF Bay Area
Manikanta Pochamalla
San Francisco Bay Area
He takes research models the last mile into production, and proves it in deltas.
Manikanta Pochamalla builds hybrid retrieval and RAG systems at Perplexity, serving 3,000+ daily users. He came up through enterprise MLOps at Accenture and a run at NVIDIA, where he fine-tuned protein and molecular foundation models — ESM2, OpenFold, MegaMolBART — for biology and drug discovery, unusual territory for a generalist engineer. A BTech in Information Technology started the thread, and a US master's carried him to the Bay Area.
- 3,000+Daily users on the Perplexity RAG platform
- −41%End-to-end latency cut with vLLM, Perplexity
- +34%Sequence-prediction accuracy at NVIDIA
HAPPENING NOW
STANDING TODAY
HOW THEY THINK3 BELIEFS
Four convictions run through every system he ships.
THE IMPACT
Every system here is scored by the delta it moved.
THE TIMELINE
He kept one stack and moved the frontier — enabling ML, then scaling it, then serving it.
The whole record, newest first – tap any moment and Aster tells the story.
2023–PRESENT
Training foundation models, then serving grounded answers
At NVIDIA he fine-tuned protein and molecular foundation models — ESM2, OpenFold, and MegaMolBART under BioNeMo — lifting sequence-prediction accuracy 34% and cutting training time 41% across multi-node GPUs. At Perplexity he now builds hybrid Vector Search + BM25 retrieval and vLLM-served RAG for 3,000+ daily users.
“I thrive in distributed systems, cloud-native architectures, and MLOps, and I enjoy driving end-to-end AI products from research and model development through deployment and production optimization.”
2021–2022
Going deep on the data substrate under ML
Finished his MS in Computer Science at Campbellsville University.
“Results-driven Data Engineer with 4+ years of experience building, optimizing, and scaling data ecosystems in cloud environments.”
2016–2020
The IT degree and enterprise ML at Accenture
The arc begins with a BTech in Information Technology at R.V.R. & J.C. College of Engineering in Andhra Pradesh. In 2019 he joined Accenture as a Machine Learning Engineer, building Python, PySpark, and Databricks pipelines that lifted model accuracy 32% and MLflow CI/CD that cut deployment time 45% for 3,000+ enterprise users.