— Hello, I'm

Arjun Bindu Jayachandran

Senior Software Engineer · Wells Fargo · Bengaluru

Building PnL microservices at scale and GenAI systems for regulated fintech.

Top Achievers Award · Wells Fargo 2025 Published · Journal of Supercomputing 2023

About

I'm a Senior Software Engineer at Wells Fargo, working at the intersection of GenAI and high-throughput backend systems for fixed-income trading. My work spans agentic LLM pipelines, neural retrieval, and distributed PnL microservices hardened for replay-safety, exactly-once Kafka semantics, and idempotent ingestion.

I joined Wells Fargo as an intern in 2021, came back full-time in 2022 right after graduating from NIT Calicut, and was promoted to Senior in 2026. Along the way I picked up a Top Achievers Award, an Innovation Pitch Day finalist spot, and an IEEE-published paper on metaheuristic feature selection.

Outside of work I'm building kharcha-ai — a multi-agent LangGraph + Gemma 4 spend-analysis system that runs entirely on a laptop — plus PyQt6 desktop AI, a focus-timer side project, local LLM experiments, and the occasional weekend hack. Scroll down for both work and personal builds.

Experience

  1. Jan 2026 – Present

    Senior Software Engineer

    Wells Fargo · Bengaluru

    • Initiated a bond recommendation system — built a two-tower dual-encoder PoC with symmetric InfoNCE contrastive learning and ANN similarity search, plus an LLM agent layer that converses with users and orchestrates similarity + free-form queries. Drove from PoC to a funded production build.
    • Architected the production deployment with decoupled training + serving (ONNX-runtime inference, FAISS vector indices, temporal train/val/test split) and feedback-driven scheduled retraining from accept/dismiss signals.
    • Shipped a replay-safe historical rerun API on a PnL service — async Camel endpoint with context-propagation headers (recalc-flag, snapshot-date) flowing through trade / position / publish routes, with date-correct persistence and Kafka-publication suppression for backfills, so corrected snapshots never leak into live downstream consumers.
    • Drove reliability hardening across PnL reporting services — led root-cause analysis on data mismatches, decoupled fragile upstream dependencies, and added fail-fast integrity guards, shifting services from patch-driven fixes to a resilient-by-design posture.
  2. Jul 2022 – Dec 2025

    Software Engineer

    Wells Fargo · Bengaluru

    • Designed a LangGraph orchestrator + validator agentic pipeline for natural-language → database query translation, layering RAG retrieval, MCP-based tool exposure, and governance (syntax checks, PII masking, tamper-evident audit) to enforce safe-query policy before execution.
    • Enabled horizontal scaling of PnL microservices — externalized in-process state to Redis to enable stateless replicas, with Kafka exactly-once consumer semantics ensuring concurrency safety under high-volume trade load. Contributed a DMN rules engine for swap position and payment-leg roll-ups and idempotent file ingestion for end-of-day archival flows.
    • Optimized MongoDB query latency by 30–40% on PnL reporting stores — replaced legacy Views with aggregation pipelines and compound indexing, materially reducing infra load across the reporting cluster.
    • Redesigned a PnL data-collection backup microservice from single-call to batched processing — consolidated per-record upstream calls into bulk payloads, improving end-of-day throughput by ~90% and reducing I/O contention across the service mesh.
    • Authored a 150+ financial-scenario automation framework (parallel execution, 60% regression-cycle cut, Splunk telemetry) and led OpenShift modernization of 10+ core services with resilient CI/CD.
    • Owned PnL services and ABF data ingestion, driving steady reliability improvements through targeted refactors and observability upgrades.
  3. May 2021 – Jul 2021

    Technology Intern

    Wells Fargo · Bengaluru

    • Built a time-series forecasting engine for ATM cash optimization, benchmarking SARIMA, LSTM, and Prophet on transaction data from 111 ATMs, with exogenous factors (holidays, weather) and automated retraining. Achieved ~91% weekly prediction accuracy via aggregated-window forecasting.

Selected work

Personal builds along the same lines as my Wells Fargo work — agentic LLM pipelines, retrieval, and applied ML — plus tooling I needed for myself. Plenty more side experiments on GitHub.

Open Source

VoiceToText-with-UI

An open-source PyQt6 AI dictation assistant. Real-time, on-device transcription via faster-whisper on NVIDIA GPUs. Hardware-level Bluetooth HFP toggling to prevent audio degradation, system-tray background execution, graceful VRAM cleanup, global hotkey, and clipboard auto-paste.

PyQt6faster-whisperCUDAWin32Whisper
View on GitHub →
Recent · Open Source

lockin · Focus Timer

A green-themed standalone Windows focus timer I built for my own interview prep. Wall-clock based (survives laptop sleep), Picture-in-Picture mini mode for always-on-top, procedurally-generated brown / pink / white / rain ambient noise via Web Audio, and File System Access API auto-export so every session streams to disk as jsonl for pandas analytics. SVG-rendered calendar heatmap, hour-of-day grid, and personal-bests dashboard. No build, no deps, ~1500 lines.

Vanilla JSWeb AudioDocument PiPFile System AccessSVGIndexedDB
GitHub →
Open Source

Lumen.AI · NL → SQL

A natural-language-to-SQL pipeline using a local Ollama sqlcoder model with a llama3 cleanup pass, MySQL execution, and a Streamlit UI. Lets non-engineers query a database in plain English with no data leaving the machine.

OllamasqlcoderLangChainStreamlitMySQL
GitHub →
Hackathon

GAIDP — Gen AI Data Profiler

Anomaly detection over Federal Reserve FR Y-14Q-style corporate-loan data using Isolation Forest on numerical fields and DBSCAN on categorical, with a token-budget-aware router that escalates flagged rows to an LLM for narrative explanation.

scikit-learntiktokenpandasJupyter
GitHub →
Led to Publication

Grey Wolf Optimizer · Feature Selection

Implementations of GWO and three variants — Binary GWO, Multi-Objective GWO, and Binary MOGWO — wrapping a Keras ANN evaluator to perform metaheuristic feature selection on UCI ML datasets. Precursor work for the SaBMOGWO-S paper above.

NumPyscikit-learnTensorFlowMetaheuristics
GitHub →
Plus more

See all on GitHub →

Older side experiments across blockchain, Linux kernel modules, Verilog FPGA, GANs, computer vision, and full-stack web — kept around as artifacts of past learning, separate from current focus.

github.com/Arjun-B-J →

Publications & Awards

Award

Top Achievers Award · Wells Fargo 2025

Wells Fargo's highest individual honor.

Award

Innovation Pitch Day 2024 · Grand Finalist

Top 4 across firm-wide submissions.

Award

Manager Spotlight Awards · ×2

Recognized for impact on the team across two separate cycles.

Skills

Primary

Daily, in production
Languages
PythonJava
Backend
Apache KafkaMongoDBRedisSpring BootApache CamelRESTgRPC
AI / Agentic
LangGraphLangChainRAGMCPTool CallingOrchestrator + Validator
Infra & CI/CD
KubernetesDockerHelmOpenShiftGitHub ActionsJenkins
Observability
SplunkGrafanaPrometheusSonar

Working knowledge

Use regularly
ML / Retrieval
PyTorchFAISSscikit-learn
Security
PII MaskingTamper-Evident Audit

From side projects

Used in personal builds
Local AI / ML
faster-whisperOllamaGemma 4TensorFlow / Keras
Web / API
FastAPINext.js 15TypeScriptTailwindReportLab
UI
PyQt6StreamlitWeb AudioDocument PiP

Education

B.Tech in Computer Science and Engineering

National Institute of Technology Calicut · 2022

CGPA 8.55

Get in touch

Open to interesting conversations — collaboration, advising, or just trading notes on agentic systems and high-throughput backends.