01
Business Challenge
Retail traders in India are left choosing between bare-bones charting apps and expensive institutional terminals, with no risk-free way to test AI-assisted research against live market data before trusting it with real capital.
AI Trading PlatformInternal Product
AkechiTrade
AkechiTrade is an AI-powered market intelligence and paper-trading platform for Indian financial markets (NSE, BSE, MCX), combining a real-time data terminal, a sandbox trading engine, and a multi-agent AI research desk under SEBI Research Analyst governance.
17
Services Shipped
5 + CIO
AI Analyst Agents
NSE·BSE·MCX
Markets Covered
SEBI RA Governed
Compliance
02
Akechi Approach
Built a full-stack platform on Azure and Azure Databricks: a real-time WebSocket market-data gateway, a virtual-money sandbox trading engine that models Indian brokerage/STT/GST charges exactly, and a LangGraph multi-agent research desk (five analysts plus a CIO) that produces daily pre-market recommendations behind an immutable audit trail and a human SEBI RA review gate.
03
Measured Outcome
Currently in active build: seventeen backend services, six Databricks lakehouse packages, and the full Next.js terminal are landing against a complete engineering documentation suite, with compliance and audit-trail guardrails built in from day one rather than retrofitted before launch.
Technical Architecture
A disciplined stack, built around the work.
The reference page feels premium because every technology has a purpose. This portfolio template does the same: grouped systems, clear roles, and compact proof.
Market Data Layer
Real-time WebSocket gateway and Redis hot path feeding candles, options Greeks, and an alerts engine off a live exchange tape.
WebSocket gatewayRedis hot pathCandles & GreeksAlerts engine
Sandbox Trading Engine
Virtual-money order lifecycle with fill simulation and the full Indian charges stack — brokerage, STT/CTT, exchange fees, GST, stamp duty.
Order lifecycleCharges engineMargin & risk checksP&L accounting
AI Research Desk
LangGraph multi-agent panel on Databricks Mosaic AI, publishing to an immutable audit trail behind a human RA review gate.
LangGraph agentsClaude on DatabricksAudit trailRA review gate
Technology Stack
Next.jsFastAPILangGraphClaudeAzure DatabricksPostgreSQLRedisTerraform
Akechi Capabilities
What we delivered, organized like an enterprise solution.
Each capability is scoped, named, and tied to the project outcome so the page reads as proof of execution, not a loose gallery.
Real-Time Terminal
Live market data, option chains, and charting for NSE, BSE, MCX, currency, and mutual funds in one Next.js terminal.
Market DataOptionsCharting
Risk-Managed Paper Trading
A sandbox engine that models real Indian order types (CNC/MIS/NRML/GTT/BO/CO) and charges against live prices with virtual money only.
Paper TradingRisk EngineCharges Model
Governed AI Recommendations
A daily pre-market pipeline where AI-generated calls are scored, audited, and gated behind SEBI Research Analyst review before publication.
LangGraphComplianceAudit Trail
Graduate-to-Live Path
A broker-bridge service behind four explicit gates — feature flag, versioned consent, step-up MFA, per-order confirmation — for users who choose to connect a real broker.
Broker BridgeDPDP ConsentKill Switch
Solution Proof
Problem, response, result without visual noise.
This mirrors the clarity of the reference page while keeping the Akechi data and brand voice intact.
Research Without Risk
Problem
Traders have no safe way to see whether AI-generated research is actually good before risking real money on it.
Akechi Response
Built a sandbox engine on live market data so every AI call and every strategy can be paper-traded first, charges and all.
Result
A track record traders can inspect before they ever fund a live account.
Compliance as Architecture
Problem
AI-generated financial content is an easy way to accidentally cross into unregistered investment advice.
Akechi Response
Wired the immutable audit trail and the human SEBI RA review gate into the publication pipeline itself, not as a policy layered on afterward.
Result
Every recommendation is traceable and reviewable by design, not by exception.
India-Specific by Default
Problem
Generic trading platforms get Indian market mechanics wrong — settlement cycles, circuit bands, MCX lot sizing, CTT vs STT.
Akechi Response
Modeled Indian market calendars, charges, and product terms (CNC/MIS/NRML/GTT) directly into the shared packages every service imports.
Result
Domain correctness that does not have to be re-derived service by service.
Delivery Framework
A controlled path from discovery to adoption.
The phase model keeps the page dense and scannable while showing clients how Akechi manages risk.
01
Specify
Write the SRD and binding architecture ADRs before code, covering compliance, data flows, and disaster recovery.
02
Build the Platform
Land shared packages, the seventeen services, and the Databricks lakehouse against that spec.
03
Wire Compliance In
Build the audit trail, SEBI RA review gate, and DPDP consent flows as load-bearing parts of the pipeline, not add-ons.
04
Verify Without a Cluster
Give every lakehouse and infra package a deterministic CI gate that checks real invariants without needing Databricks or Terraform applied.
05
Go Live
Work the go-live runbook — credentials, registrations, and the day-two rotation calendar — toward a supervised launch.
AI & Automation Highlights
Smart systems, measured by operational impact.
This section gives the Azure-style technical confidence without changing the Akechi palette or overloading the page with decoration.
Deterministic Compliance Gates
Each lakehouse package ships a pure-Python CI gate that checks its Databricks-reference semantics without a live cluster.
7 CI gates
IaC Verified Before Apply
Terraform, Helm, and GitHub Actions are asserted against parsed files — naming, private access, deploy windows — before anything is ever applied.
Zero live applies
Audit-First Publication
Every AI recommendation writes to the immutable audit trail before it can be published, never after.
100% pre-publish audit
Ready for a similar build?
Building something that needs both AI research and real compliance guardrails?
Akechi Webcraft can design and build platforms where AI, trading logic, and regulatory constraints have to work together from day one.