diff --git a/brokers/redis/broker.go b/brokers/redis/broker.go
index 2bf3c69..6603293 100644
--- a/brokers/redis/broker.go
+++ b/brokers/redis/broker.go
@@ -29,18 +29,19 @@ type Options struct {
 }
 
 type Broker struct {
-	log        *slog.Logger
-	conn       redis.UniversalClient
-	pollPeriod time.Duration
+	log  *slog.Logger
+	opts Options
+
+	conn redis.UniversalClient
 }
 
 func New(o Options, lo *slog.Logger) *Broker {
-	pollPeriod := o.PollPeriod
 	if o.PollPeriod == 0 {
-		pollPeriod = DefaultPollPeriod
+		o.PollPeriod = DefaultPollPeriod
 	}
 	return &Broker{
-		log: lo,
+		opts: o,
+		log:  lo,
 		conn: redis.NewUniversalClient(&redis.UniversalOptions{
 			Addrs:        o.Addrs,
 			DB:           o.DB,
@@ -51,7 +52,6 @@ func New(o Options, lo *slog.Logger) *Broker {
 			MinIdleConns: o.MinIdleConns,
 			IdleTimeout:  o.IdleTimeout,
 		}),
-		pollPeriod: pollPeriod,
 	}
 }
 
@@ -87,7 +87,7 @@ func (b *Broker) Consume(ctx context.Context, work chan []byte, queue string) {
 			return
 		default:
 			b.log.Debug("receiving from consumer..")
-			res, err := b.conn.BLPop(ctx, b.pollPeriod, queue).Result()
+			res, err := b.conn.BLPop(ctx, b.opts.PollPeriod, queue).Result()
 			if err != nil && err.Error() != "redis: nil" {
 				b.log.Error("error consuming from redis queue", "error", err)
 			} else if errors.Is(err, redis.Nil) {
@@ -105,7 +105,7 @@ func (b *Broker) Consume(ctx context.Context, work chan []byte, queue string) {
 }
 
 func (b *Broker) consumeScheduled(ctx context.Context, queue string) {
-	poll := time.NewTicker(b.pollPeriod)
+	poll := time.NewTicker(b.opts.PollPeriod)
 
 	for {
 		select {
diff --git a/brokers/redis/broker_piped.go b/brokers/redis/broker_piped.go
new file mode 100644
index 0000000..fc0d5b4
--- /dev/null
+++ b/brokers/redis/broker_piped.go
@@ -0,0 +1,176 @@
+package redis
+
+import (
+	"context"
+	"errors"
+	"fmt"
+	"log/slog"
+	"strconv"
+	"time"
+
+	"github.com/go-redis/redis/v8"
+)
+
+const (
+	DefaultPipePeriod = 200 * time.Millisecond
+)
+
+type PipeBroker struct {
+	log  *slog.Logger
+	opts PipedOptions
+
+	conn redis.UniversalClient
+	pipe redis.Pipeliner
+}
+
+type PipedOptions struct {
+	Addrs        []string
+	Password     string
+	DB           int
+	DialTimeout  time.Duration
+	ReadTimeout  time.Duration
+	WriteTimeout time.Duration
+	IdleTimeout  time.Duration
+	MinIdleConns int
+
+	PollPeriod time.Duration
+	PipePeriod time.Duration
+}
+
+func NewPiped(o PipedOptions, lo *slog.Logger) *PipeBroker {
+	if o.PollPeriod == 0 {
+		o.PollPeriod = DefaultPollPeriod
+	}
+	if o.PipePeriod == 0 {
+		o.PipePeriod = DefaultPipePeriod
+	}
+
+	conn := redis.NewUniversalClient(&redis.UniversalOptions{
+		Addrs:        o.Addrs,
+		DB:           o.DB,
+		Password:     o.Password,
+		DialTimeout:  o.DialTimeout,
+		ReadTimeout:  o.ReadTimeout,
+		WriteTimeout: o.WriteTimeout,
+		MinIdleConns: o.MinIdleConns,
+		IdleTimeout:  o.IdleTimeout,
+	})
+
+	p := &PipeBroker{
+		log:  lo,
+		conn: conn,
+		pipe: conn.Pipeline(),
+		opts: o,
+	}
+
+	go p.pushPipe(context.TODO())
+
+	return p
+}
+
+func (r *PipeBroker) pushPipe(ctx context.Context) {
+	tk := time.NewTicker(r.opts.PipePeriod)
+	for {
+		select {
+		case <-ctx.Done():
+			return
+		case <-tk.C:
+			r.log.Debug("submitting redis pipe")
+			if r.pipe.Len() == 0 {
+				continue
+			}
+			if _, err := r.pipe.Exec(ctx); err != nil {
+				r.log.Error("error executing redis pipe: %v", err)
+			}
+		}
+	}
+}
+
+func (r *PipeBroker) GetPending(ctx context.Context, queue string) ([]string, error) {
+	rs, err := r.conn.LRange(ctx, queue, 0, -1).Result()
+	if err == redis.Nil {
+		return []string{}, nil
+	} else if err != nil {
+		return []string{}, err
+	}
+
+	return rs, nil
+}
+
+func (b *PipeBroker) Enqueue(ctx context.Context, msg []byte, queue string) error {
+	return b.pipe.LPush(ctx, queue, msg).Err()
+}
+
+func (b *PipeBroker) EnqueueScheduled(ctx context.Context, msg []byte, queue string, ts time.Time) error {
+	return b.pipe.ZAdd(ctx, fmt.Sprintf(sortedSetKey, queue), &redis.Z{
+		Score:  float64(ts.UnixNano()),
+		Member: msg,
+	}).Err()
+}
+
+func (b *PipeBroker) Consume(ctx context.Context, work chan []byte, queue string) {
+	go b.consumeScheduled(ctx, queue)
+
+	for {
+		select {
+		case <-ctx.Done():
+			b.log.Debug("shutting down consumer..")
+			return
+		default:
+			b.log.Debug("receiving from consumer..")
+			res, err := b.conn.BLPop(ctx, b.opts.PollPeriod, queue).Result()
+			if err != nil && err.Error() != "redis: nil" {
+				b.log.Error("error consuming from redis queue", "error", err)
+			} else if errors.Is(err, redis.Nil) {
+				b.log.Debug("no tasks to consume..", "queue", queue)
+			} else {
+				msg, err := blpopResult(res)
+				if err != nil {
+					b.log.Error("error parsing response from redis", "error", err)
+					return
+				}
+				work <- []byte(msg)
+			}
+		}
+	}
+}
+
+func (b *PipeBroker) consumeScheduled(ctx context.Context, queue string) {
+	poll := time.NewTicker(b.opts.PollPeriod)
+
+	for {
+		select {
+		case <-ctx.Done():
+			b.log.Debug("shutting down scheduled consumer..")
+			return
+		case <-poll.C:
+			b.conn.Watch(ctx, func(tx *redis.Tx) error {
+				// Fetch the tasks with score less than current time. These tasks have been scheduled
+				// to be queued.
+				tasks, err := tx.ZRevRangeByScore(ctx, fmt.Sprintf(sortedSetKey, queue), &redis.ZRangeBy{
+					Min:    "0",
+					Max:    strconv.FormatInt(time.Now().UnixNano(), 10),
+					Offset: 0,
+					Count:  1,
+				}).Result()
+				if err != nil {
+					return err
+				}
+
+				for _, task := range tasks {
+					if err := b.Enqueue(ctx, []byte(task), queue); err != nil {
+						return err
+					}
+				}
+
+				// Remove the tasks
+				if err := tx.ZRem(ctx, fmt.Sprintf(sortedSetKey, queue), tasks).Err(); err != nil {
+					return err
+				}
+
+				return nil
+			})
+		}
+
+	}
+}
diff --git a/results/redis/results_piped.go b/results/redis/results_piped.go
new file mode 100644
index 0000000..00446e7
--- /dev/null
+++ b/results/redis/results_piped.go
@@ -0,0 +1,202 @@
+package redis
+
+import (
+	"context"
+	"log/slog"
+	"strconv"
+	"time"
+
+	"github.com/go-redis/redis/v8"
+)
+
+const DefaultPipePeriod = 200 * time.Millisecond
+
+type PipedResults struct {
+	lo  *slog.Logger
+	opt PipedOptions
+
+	conn redis.UniversalClient
+	pipe redis.Pipeliner
+}
+
+type PipedOptions struct {
+	Addrs        []string
+	Password     string
+	DB           int
+	DialTimeout  time.Duration
+	ReadTimeout  time.Duration
+	WriteTimeout time.Duration
+	IdleTimeout  time.Duration
+	Expiry       time.Duration
+	MetaExpiry   time.Duration
+	MinIdleConns int
+
+	PipePeriod time.Duration
+}
+
+func NewPiped(o PipedOptions, lo *slog.Logger) *PipedResults {
+	if o.PipePeriod == 0 {
+		o.PipePeriod = DefaultPipePeriod
+	}
+
+	conn := redis.NewUniversalClient(&redis.UniversalOptions{
+		Addrs:        o.Addrs,
+		DB:           o.DB,
+		Password:     o.Password,
+		DialTimeout:  o.DialTimeout,
+		ReadTimeout:  o.ReadTimeout,
+		WriteTimeout: o.WriteTimeout,
+		MinIdleConns: o.MinIdleConns,
+		IdleTimeout:  o.IdleTimeout,
+	})
+
+	p := &PipedResults{
+		lo:   lo,
+		conn: conn,
+		pipe: conn.Pipeline(),
+		opt:  o,
+	}
+
+	// TODO: pass ctx here somehow
+	if o.MetaExpiry != 0 {
+		go p.expireMeta(context.TODO(), o.MetaExpiry)
+	}
+
+	go p.execPipe(context.TODO())
+
+	return p
+}
+
+func (r *PipedResults) execPipe(ctx context.Context) {
+	tk := time.NewTicker(r.opt.PipePeriod)
+	for {
+		select {
+		case <-ctx.Done():
+			r.lo.Info("context closed, draining redis pipe", "length", r.pipe.Len())
+			if _, err := r.pipe.Exec(ctx); err != nil {
+				r.lo.Error("error executing redis pipe: %v", err)
+			}
+			return
+		case <-tk.C:
+			plen := r.pipe.Len()
+			if plen == 0 {
+				continue
+			}
+
+			r.lo.Info("submitting redis pipe", "length", plen)
+			if _, err := r.pipe.Exec(ctx); err != nil {
+				r.lo.Error("error executing redis pipe: %v", err)
+			}
+		}
+	}
+}
+
+func (r *PipedResults) DeleteJob(ctx context.Context, id string) error {
+	r.lo.Debug("deleting job")
+	if err := r.conn.ZRem(ctx, resultPrefix+success, 1, id).Err(); err != nil {
+		return err
+	}
+
+	if err := r.conn.ZRem(ctx, resultPrefix+failed, 1, id).Err(); err != nil {
+		return err
+	}
+
+	if err := r.conn.Del(ctx, resultPrefix+id).Err(); err != nil {
+		return err
+	}
+
+	return nil
+}
+
+func (r *PipedResults) GetSuccess(ctx context.Context) ([]string, error) {
+	// Fetch the failed tasks with score less than current time
+	r.lo.Debug("getting successful jobs")
+	rs, err := r.conn.ZRevRangeByScore(ctx, resultPrefix+success, &redis.ZRangeBy{
+		Min: "0",
+		Max: strconv.FormatInt(time.Now().UnixNano(), 10),
+	}).Result()
+	if err != nil {
+		return nil, err
+	}
+
+	return rs, nil
+}
+
+func (r *PipedResults) GetFailed(ctx context.Context) ([]string, error) {
+	// Fetch the failed tasks with score less than current time
+	r.lo.Debug("getting failed jobs")
+	rs, err := r.conn.ZRevRangeByScore(ctx, resultPrefix+failed, &redis.ZRangeBy{
+		Min: "0",
+		Max: strconv.FormatInt(time.Now().UnixNano(), 10),
+	}).Result()
+	if err != nil {
+		return nil, err
+	}
+
+	return rs, nil
+}
+
+func (r *PipedResults) SetSuccess(ctx context.Context, id string) error {
+	r.lo.Debug("setting job as successful", "id", id)
+	return r.pipe.ZAdd(ctx, resultPrefix+success, &redis.Z{
+		Score:  float64(time.Now().UnixNano()),
+		Member: id,
+	}).Err()
+}
+
+func (r *PipedResults) SetFailed(ctx context.Context, id string) error {
+	r.lo.Debug("setting job as failed", "id", id)
+	return r.pipe.ZAdd(ctx, resultPrefix+failed, &redis.Z{
+		Score:  float64(time.Now().UnixNano()),
+		Member: id,
+	}).Err()
+}
+
+func (r *PipedResults) Set(ctx context.Context, id string, b []byte) error {
+	r.lo.Debug("setting result for job", "id", id)
+	return r.pipe.Set(ctx, resultPrefix+id, b, r.opt.Expiry).Err()
+}
+
+func (r *PipedResults) Get(ctx context.Context, id string) ([]byte, error) {
+	r.lo.Debug("getting result for job", "id", id)
+	rs, err := r.conn.Get(ctx, resultPrefix+id).Bytes()
+	if err != nil {
+		return nil, err
+	}
+
+	return rs, nil
+}
+
+// TODO: accpet a ctx here and shutdown gracefully
+func (r *PipedResults) expireMeta(ctx context.Context, ttl time.Duration) {
+	r.lo.Info("starting results meta purger", "ttl", ttl)
+
+	var (
+		tk = time.NewTicker(ttl)
+	)
+
+	for {
+		select {
+		case <-ctx.Done():
+			r.lo.Info("shutting down meta purger", "ttl", ttl)
+			return
+		case <-tk.C:
+			now := time.Now().UnixNano() - int64(ttl)
+			score := strconv.FormatInt(now, 10)
+
+			r.lo.Debug("purging failed results metadata", "score", score)
+			if err := r.pipe.ZRemRangeByScore(context.Background(), resultPrefix+failed, "0", score).Err(); err != nil {
+				r.lo.Error("could not expire success/failed metadata", "err", err)
+			}
+
+			r.lo.Debug("purging success results metadata", "score", score)
+			if err := r.pipe.ZRemRangeByScore(context.Background(), resultPrefix+success, "0", score).Err(); err != nil {
+				r.lo.Error("could not expire success/failed metadata", "err", err)
+			}
+		}
+	}
+}
+
+func (r *PipedResults) NilError() error {
+	return redis.Nil
+}