Agent Engineering
ASURA - Autonomous CLI Agent
A structured AI planning engine that translates natural language instructions into validated filesystem actions and shell commands - with built-in safety guardrails, memory, and reversible execution.
Architectural Focus
Designed to explore reliable agent behavior by enforcing schema validation, retry logic, command risk analysis, and contextual memory - reducing hallucinated execution while maintaining developer control.
• LLM → structured execution planner
• Zod-validated action schemas
• Dangerous command detection
• Stateful memory layer
• Clarification-first safety model
• Undo-aware file operations
> asura "create bang server" ✔ intent: create ✔ files generated ✔ commands prepared ⚠ risk: low Executing...
Featured Engineering Work
BangJS - Runtime-Oriented Node Framework
An exploration into predictable middleware execution, modular routing, and schema-driven validation - built to better understand backend runtime mechanics rather than abstract them away.
Why Build Another Framework?
Many backend frameworks optimize for convenience but obscure execution flow. BangJS was built to make middleware order, validation boundaries, and request lifecycles explicit - improving predictability during development and debugging.
• Execution-first middleware pipeline
• Router abstraction for modular scaling
• Route-level schema validation
• Decoupled response handling
• Built from scratch to understand runtime constraints
// schema validation
app.post('/check', handler, {
age: { required: true }
});Product Engineering
XAR - Zero-Download Archive Inspector
Built to eliminate the friction of downloading compressed files, XAR enables users to explore archive contents instantly while providing behavioral analytics for deeper usage insight.
Engineering Focus
XAR reconstructs archive structure in-memory, allowing users to preview files across multiple formats without extraction. The system was designed with a strong emphasis on usability, observability, and efficient client-side processing.
• Remote ZIP parsing without forced downloads
• Dynamic filesystem tree generation
• Multi-format preview engine
• Event-driven analytics tracking
• Metadata-aware archive inspection
// remote archive parsing
const blob = await fetch(zipUrl).then(r => r.blob()); const zip = await JSZip.loadAsync(blob); const tree = buildTree(zip);
Systems Architecture
RabbitHole - Concept Graph Learning Engine
RabbitHole models knowledge as a traversable graph, enabling learners to move through concepts via structured pathways such as deeper exploration, lateral expansion, and failure-driven insight. Designed as a backend-first system, it emphasizes domain clarity, progression logic, and cognitive load management.
Engineering Focus
The system treats concepts as nodes within a directed graph, allowing structured traversal based on learner intent. Backend services orchestrate doorway resolution while maintaining progression state and difficulty awareness.
• Graph-based domain modeling
• Doorway-driven concept traversal
• Difficulty-aware progression engine
• Lifecycle-managed knowledge states
• Backend-first architecture
// doorway resolution
@Document(collection = "concepts")
public class Concept {
@Id
private String id;
private String title;
private Difficulty difficulty;
private List<Doorway> doorways;
}Beyond Production
Telugitha - A Modern Telugu Programming Language
+A PEG.js-based language experiment exploring grammar construction and interpreter-style execution.
Touchless Interface Controller
+A computer vision system that translates hand gestures into real-time operating system controls.
Operational Experience
Trusted Near Production Systems
Accenture - Associate Software Engineer
Oct 2025 - Present
Previously Software Engineer Intern
Feb 2025 - Jun 2025
Worked inside enterprise HR infrastructure supporting hiring pipelines, release confidence, and workflow reliability across production environments.
• Validated offer-to-onboarding workflows impacting real employee lifecycles
• Operated within structured deployment cycles and cross-team dependencies
• Strengthened system reliability through disciplined testing practices
• Gained exposure to large-scale operational behavior and enterprise constraints