feat: Intelligence Pipeline v3 — full implementation

Multi-stage evidence-grounded inference architecture replacing the monolithic 9B model extraction pipeline. CPU-first specialist services handle routine extraction while the 9B vLLM model is preserved for semantic adjudication of ambiguous cases. Key components: - Capability-aware inference gateway (OpenAI-compatible + Ollama) - Endpoint registry with DB migrations and REST API - Sentence-aware document segmenter (property tests) - Deterministic financial parsing with offset integrity - Symbol resolution with ambiguity detection - Specialist service (GLiNER2, dynamic batching, K8s deployment) - Company-specific sentiment (FinBERT, calibration) - Retrieval-based novelty and duplicate detection - Confidence calibration pipeline - Deterministic routing engine (property tests) - 9B adjudication layer with VRAM gating - Stock-specific impact model (features, labels, baseline, trained) - Pipeline orchestrator (state machine, queues, leases, feature flags) - Bounded parallelism (async workers, semaphore, load shedding) - Observability (tracing, metrics, alerts) - Compatibility adapter (v3→v2 golden mapping tests) - Shadow/canary promotion framework - Active learning and fine-tuning pipeline Test results: 1,161 tests pass, ruff lint clean. All 282 spec tasks completed.
2026-07-13 02:14:59 +00:00
parent 84634a365e
commit a72f336ad1
227 changed files with 50403 additions and 0 deletions
@@ -0,0 +1,108 @@
+{{- if .Values.specialist }}
+{{- if .Values.specialist.enabled }}
+---
+apiVersion: apps/v1
+kind: Deployment
+metadata:
+  name: specialist
+  namespace: {{ .Release.Namespace }}
+  labels:
+    app: specialist
+    {{- include "stonks.labels" . | nindent 4 }}
+    stonks-oracle/tier: processing
+spec:
+  replicas: {{ .Values.specialist.replicas | default 2 }}
+  selector:
+    matchLabels:
+      app: specialist
+  template:
+    metadata:
+      labels:
+        app: specialist
+        stonks-oracle/tier: processing
+    spec:
+      automountServiceAccountToken: false
+      {{- with .Values.imagePullSecrets }}
+      imagePullSecrets:
+        {{- toYaml . | nindent 8 }}
+      {{- end }}
+      securityContext:
+        {{- include "stonks.podSecurityContext" . | nindent 8 }}
+      containers:
+        - name: specialist
+          image: {{ .Values.image.registry }}/specialist:{{ .Values.image.tag }}
+          imagePullPolicy: {{ .Values.image.pullPolicy }}
+          command: ["sh", "-c", "uvicorn services.specialist.app:app --host 0.0.0.0 --port 8000"]
+          ports:
+            - containerPort: 8000
+          env:
+            - name: SPECIALIST_MODEL
+              value: {{ .Values.specialist.model | default "urchade/gliner_large-v2.1" | quote }}
+            - name: SPECIALIST_MAX_BATCH_SIZE
+              value: {{ .Values.specialist.maxBatchSize | default "32" | quote }}
+            - name: SPECIALIST_MAX_WAIT_MS
+              value: {{ .Values.specialist.maxWaitMs | default "50.0" | quote }}
+            - name: SPECIALIST_MAX_QUEUE_SIZE
+              value: {{ .Values.specialist.maxQueueSize | default "256" | quote }}
+            - name: SPECIALIST_TEST_MODE
+              value: {{ .Values.specialist.testMode | default "0" | quote }}
+          securityContext:
+            {{- include "stonks.containerSecurityContext" . | nindent 12 }}
+          envFrom:
+            - configMapRef:
+                name: stonks-config
+            {{- range .Values.specialist.secrets }}
+            - secretRef:
+                name: {{ . }}
+            {{- end }}
+          resources:
+            requests:
+              cpu: {{ .Values.specialist.resources.requests.cpu | default "2" | quote }}
+              memory: {{ .Values.specialist.resources.requests.memory | default "4Gi" }}
+            limits:
+              cpu: {{ .Values.specialist.resources.limits.cpu | default "6" | quote }}
+              memory: {{ .Values.specialist.resources.limits.memory | default "10Gi" }}
+          readinessProbe:
+            httpGet:
+              path: /ready
+              port: 8000
+            initialDelaySeconds: 10
+            periodSeconds: 10
+            timeoutSeconds: 5
+          livenessProbe:
+            httpGet:
+              path: /health
+              port: 8000
+            initialDelaySeconds: 30
+            periodSeconds: 30
+            timeoutSeconds: 5
+          volumeMounts:
+            - name: tmp
+              mountPath: /tmp
+            - name: model-cache
+              mountPath: /root/.cache
+      volumes:
+        - name: tmp
+          emptyDir:
+            sizeLimit: 10Mi
+        - name: model-cache
+          emptyDir:
+            sizeLimit: 5Gi
+---
+apiVersion: v1
+kind: Service
+metadata:
+  name: specialist
+  namespace: {{ .Release.Namespace }}
+  labels:
+    app: specialist
+    {{- include "stonks.labels" . | nindent 4 }}
+spec:
+  selector:
+    app: specialist
+  ports:
+    - port: 8000
+      targetPort: 8000
+      protocol: TCP
+{{- end }}
+{{- end }}
@@ -289,6 +289,24 @@ superset:
    requests: { cpu: 200m, memory: 512Mi }
    limits: { cpu: "1", memory: 2Gi }

+## Specialist inference service (CPU-first NER/classification)
+specialist:
+  enabled: true
+  replicas: 2
+  model: "urchade/gliner_large-v2.1"
+  maxBatchSize: "32"
+  maxWaitMs: "50.0"
+  maxQueueSize: "256"
+  testMode: "0"
+  secrets: [stonks-core-secrets]
+  resources:
+    requests:
+      cpu: "2"
+      memory: 4Gi
+    limits:
+      cpu: "6"
+      memory: 10Gi
+
 ## Network policies
 networkPolicies:
  enabled: true