I work where quantum computing, agentic AI, and production software collide: QUBO and Ising formulations, QAOA and hybrid solvers, LangGraph agents, RAG pipelines, FastAPI services, Next.js product surfaces, Dockerized deployments, and evaluation loops that make the system better over time.
Research depth. Clean architecture. Shipping velocity.
class KondapiSriPranav:
location = "Vijayawada, India"
role = "Quantum Computing x Agentic AI Engineer"
builds = [
"hybrid quantum-classical optimizers",
"agentic AI products with real tool use",
"RAG systems with evaluation and citations",
"typed APIs, dashboards, and deployment pipelines",
]
stack = {
"quantum": ["Qiskit", "QAOA", "VQE", "QUBO", "Ising"],
"ai": ["LangGraph", "LangChain", "RAG", "PyTorch", "Hugging Face"],
"systems": ["Python", "FastAPI", "Next.js", "PostgreSQL", "Docker"],
}
def north_star(self):
return "Make advanced ideas useful, testable, and shippable." |
Currently focused on Hybrid solvers, agent evaluation, quantum optimization, MCP-style tool workflows, and product-grade AI infrastructure. |
One pattern that holds my whole stack together — research formulation, quantum execution, classical optimization, and the eval loop that decides if any of it actually worked.
def hybrid_loop(problem, max_iter: int = 100):
"""A pattern I keep coming back to: hybrid quantum-classical optimization."""
qubo = formulate(problem) # 1. research — turn the world into a Hamiltonian
qaoa = QAOA(cost_hamiltonian=qubo, p=3)# 2. quantum — variational ansatz
metric = Evaluator(problem) # 3. eval — observable, not vibes
for step in range(max_iter):
params = classical_optimizer(qaoa) # 4. classical — gradient / COBYLA / SPSA
bitstr = qaoa.run(params) # 5. execute — circuit on simulator or hardware
score = metric(bitstr) # measure — every loop must be defensible
if metric.converged(score):
return Result(bitstr=bitstr, score=score, history=metric.history)
return metric.best_so_far()
|
|
|
|
|
Hybrid quantum-classical portfolio optimizer. Mean-variance with cardinality posed as QUBO, run via QAOA against a classical baseline for honest benchmarking.
|
Plans entanglement distribution across quantum networks with explicit fidelity budgeting. Research-grade tooling for repeater placement and route selection.
|
|
Multi-channel customer-query router built around MCP-style tool specs. Routes intents across channels with structured tool calls and explicit fallback paths.
|
Agentic CLI with tool use, persistent memory, and shell-aware reasoning. Built for engineers who live in the terminal.
|
|
Document intelligence with hybrid retrieval, reranking, and citations. Source-grounded answers from your PDFs — eval-ready and audit-friendly.
|
Languages
AI, ML, and Data
Quantum
Product Engineering
More tools I reach for
Poetry · uv · pytest · ruff · black · mypy · Vitest · OpenTelemetry · Prometheus · Grafana · Sentry · pandas · NumPy · Polars · DuckDB · pgvector · FAISS
What I'm reading right now
|
Quantum
|
AI Systems
|
Engineering
|
flowchart LR
A["Research problem"] --> B["Mathematical formulation"]
B --> C["Baseline implementation"]
C --> D["Evaluation harness"]
D --> E["Product API"]
E --> F["Interface"]
F --> G["Deployment"]
G --> H["Measure, debug, iterate"]
H --> D
The loop on the right is the part most projects skip. I treat it as the product.
|
I like paper-to-product work: understand the math, extract the usable mechanism, and prove it with experiments. |
I care about APIs, contracts, observability, failure modes, and code that another engineer can inherit. |
I build end-to-end: model, backend, frontend, deployment, docs, and the feedback loop after launch. |
|
Hybrid solver benchmarks and optimization copilots. |
QAOA depth, noise behavior, and error mitigation. |
Agent workflows with measurable tool use. |
Quantum, AI, research engineering, and serious product work. |
If you are working on quantum optimization, agentic AI, intelligent infrastructure, or a research idea that deserves production execution, I want to hear about it.
I usually reply within 24 hours.
Designed with intent. Built to ship. Tuned for signal.